Network fingerprint mining model training method, device, storage medium and electronic equipment

Abstract

The present application discloses a network fingerprint mining model training method, device, storage medium and electronic device, wherein the method includes: obtaining a first sample data set corresponding to a transaction party and a real sample score corresponding to each first sample data, training a pre-created first network fingerprint mining model based on the first sample data set and the real sample score corresponding to each first sample data, obtaining a second network fingerprint mining model after the first network fingerprint mining model training converges, inputting the first sample data set into the second network fingerprint mining model, obtaining at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data, training a pre-created third network fingerprint mining model based on the second positive sample data cluster and the second prediction score corresponding to each first sample data, and obtaining a fourth network fingerprint mining model after the training converges.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a network fingerprint mining model training method, device, storage medium and electronic device. Background Art

[0002] With the booming development of the service industry, the delivery service industry has also developed rapidly. Delivery service means that the business party issues a delivery order, and the delivery party rushes to the business party according to the pickup location recorded in the delivery order to pick up the delivery object corresponding to the delivery order, and then delivers the delivery object corresponding to the delivery order to the location specified in the delivery order and delivers it to the customer. In recent years, with the increasing number of business parties, the reasonable scheduling of delivery parties to meet the delivery needs of the business parties has become one of the challenges facing the delivery service industry. Reasonable scheduling of delivery parties requires accurate positioning of the delivery parties. Among the relevant technologies, for outdoor environments, precise positioning can be achieved based on satellite positioning systems, and for indoor environments, positioning can be achieved based on Wi-Fi fingerprints.

[0003] However, Wi-Fi fingerprint-based positioning technology requires fixed-point Wi-Fi signal collection of the transaction party to be located. As the number of transaction parties increases, manual collection of Wi-Fi fingerprint features near the transaction parties is inefficient and costly. Summary of the Invention

[0004] The embodiments of the present application provide a network fingerprint mining model training method, device, storage medium, and electronic device that can solve the problem of low efficiency and high cost of manually collecting network fingerprint features near the transaction party. The technical solution is as follows:

[0005] In a first aspect, an embodiment of the present application provides a network fingerprint mining model training method, the method comprising:

[0006] Obtaining a first sample data set corresponding to a transaction party and a true sample score corresponding to each first sample data point, wherein the first sample data includes a first network data point, a positive sample label or a negative sample label corresponding to the first network data point, and delivery attribute information corresponding to the first network data point, wherein the first network data point includes at least one wireless network;

[0007] Training a pre-created first network fingerprint mining model based on the first sample data set and the real sample scores corresponding to each of the first sample data to obtain a second network fingerprint mining model after the training of the first network fingerprint mining model converges;

[0008] Inputting the first sample data set into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster;

[0009] The pre-created third network fingerprint mining model is trained based on the second positive sample data cluster and the second prediction scores corresponding to each of the first sample data in the second positive sample data cluster to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges. The model structure and model parameters of the third network fingerprint mining model and the second network fingerprint mining model are the same.

[0010] In a second aspect, an embodiment of the present application provides a network fingerprint mining method, the method comprising:

[0011] Acquire a third sample data set corresponding to the target transaction party, the third sample data including a third network data point and delivery attribute information corresponding to the third network data point, the third network data point including at least one wireless network;

[0012] The third sample data set is input into the fourth network fingerprint mining model that has been trained using the network fingerprint mining model training method according to any one of claims 1 to 6 to obtain the network fingerprint feature corresponding to the target transaction party, wherein the network fingerprint feature includes at least one target positive sample data cluster and a target positive sample score threshold, and the network fingerprint feature is used to determine whether the target network data point reported by the delivery party when executing the delivery task issued by the target transaction party can be used as the network fingerprint corresponding to the target transaction party.

[0013] In a third aspect, an embodiment of the present application provides a network fingerprint mining model training device, the device comprising:

[0014] A first sample acquisition module is configured to acquire a first sample data set corresponding to a transaction party and a true sample score corresponding to each first sample data point, wherein the first sample data includes a first network data point, a positive sample label or a negative sample label corresponding to the first network data point, and delivery attribute information corresponding to the first network data point, wherein the first network data point includes at least one wireless network;

[0015] A first model training module is configured to train a pre-created first network fingerprint mining model based on the first sample data set and the real sample scores corresponding to each of the first sample data, to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges;

[0016] A second sample acquisition module is configured to input the first sample data set into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster;

[0017] The second model training module is used to train the pre-created third network fingerprint mining model based on the second positive sample data cluster and the second prediction scores corresponding to each of the first sample data in the second positive sample data cluster, to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges, and the model structure and model parameters of the third network fingerprint mining model and the second network fingerprint mining model are the same.

[0018] In a fourth aspect, an embodiment of the present application provides a network fingerprint mining device, the device comprising:

[0019] A third sample acquisition module is configured to acquire a third sample data set corresponding to the target transaction party, wherein the third sample data includes a third network data point and delivery attribute information corresponding to the third network data point, wherein the third network data point includes at least one wireless network;

[0020] A fingerprint feature determination module is used to input the third sample data set into a fourth network fingerprint mining model that has been trained using the network fingerprint mining model training method as described above, to obtain the network fingerprint feature corresponding to the target transaction party, wherein the network fingerprint feature includes at least one target positive sample data cluster and a target positive sample score threshold. The network fingerprint feature is used to determine whether the target network data point reported by the delivery party when executing the delivery task issued by the target transaction party can be used as the network fingerprint corresponding to the target transaction party.

[0021] In a fifth aspect, an embodiment of the present application provides a storage medium, wherein the storage medium stores at least one instruction, and the at least one instruction is suitable for being loaded by a processor and executing the above-mentioned method steps.

[0022] In a sixth aspect, an embodiment of the present application provides an electronic device, which may include: a processor and a memory; wherein the memory stores at least one instruction, and the at least one instruction is suitable for being loaded by the processor and executing the above-mentioned method steps.

[0023] The beneficial effects of the technical solutions provided by some embodiments of the present application include at least:

[0024] In an embodiment of the present application, a first sample data set corresponding to a transaction party and a real sample score corresponding to each first sample data are first obtained, and then a pre-created first network fingerprint mining model is trained based on the first sample data set and the real sample score corresponding to each first sample data to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges. The first sample data set is then input into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster. Finally, a pre-created third network fingerprint mining model is trained based on the second positive sample data cluster and the second prediction score corresponding to each first sample data in the second positive sample data cluster to obtain a fourth network fingerprint mining model after the third network fingerprint mining model training converges. By using the embodiment of the present application, a network fingerprint mining model that can predict the network fingerprint features corresponding to the transaction party without manual collection can be finally obtained, thereby solving the problem of low efficiency and high cost of manually collecting network fingerprint features near the transaction party. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0026] Figure 1 A flowchart of a network fingerprint mining model training method provided in an embodiment of the present application;

[0027] Figure 2 A schematic diagram of a scenario for reporting network data points provided in an embodiment of the present application;

[0028] Figure 3 A flowchart of a network fingerprint mining model training method provided in an embodiment of the present application;

[0029] Figure 4 A flowchart of a network fingerprint mining model training method provided in an embodiment of the present application;

[0030] Figure 5 A flowchart of a network fingerprint mining model training method provided in an embodiment of the present application;

[0031] Figure 6 A flowchart of a network fingerprint mining model training method provided in an embodiment of the present application;

[0032] Figure 7A flowchart of a network fingerprint mining method provided in an embodiment of the present application;

[0033] Figure 8 A flowchart of a network fingerprint mining method provided in an embodiment of the present application;

[0034] Figure 9 A schematic diagram of the structure of a network fingerprint mining model training device provided in an embodiment of the present application;

[0035] Figure 10 A schematic diagram of the structure of a network fingerprint mining device provided in an embodiment of the present application;

[0036] Figure 11 A structural block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0038] In the description of this application, it should be understood that the terms "first", "second", etc. are used for descriptive purposes only and should not be understood to indicate or imply relative importance. In the description of this application, it should be noted that, unless otherwise expressly specified and limited, "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or devices. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to the specific circumstances. In addition, in the description of this application, unless otherwise specified, "multiple" refers to two or more. "and / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0039] Indoors, GPS signals are extremely weak or even undetectable due to signal obstruction by buildings, making them unusable for positioning services. Indoor positioning technology is becoming an increasingly essential service in our lives. With the widespread adoption and deployment of Wi-Fi technology, positioning technology based on Wi-Fi fingerprinting has become the most universal and fundamental indoor positioning solution. However, for Wi-Fi fingerprint positioning technology, the coverage and freshness of the collected fingerprints are the most important factors determining positioning accuracy.

[0040] In related technologies, the Wi-Fi fingerprints of indoor merchants need to be collected manually, which cannot meet the needs of instant delivery industries such as food delivery and express delivery.

[0041] Based on this, an embodiment of the present application proposes a network fingerprint mining model training method, which first obtains a first sample data set corresponding to the transaction party and a real sample score corresponding to each first sample data, then trains a pre-created first network fingerprint mining model based on the first sample data set and the real sample score corresponding to each first sample data, and obtains a second network fingerprint mining model after the first network fingerprint mining model training converges, then inputs the first sample data set into the second network fingerprint mining model, obtains at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster, and finally trains a pre-created third network fingerprint mining model based on the second positive sample data cluster and the second prediction score corresponding to each first sample data in the second positive sample data cluster, and obtains a fourth network fingerprint mining model after the third network fingerprint mining model training converges; using the embodiment of the present application, a network fingerprint mining model can be finally obtained that can predict the network fingerprint features corresponding to the transaction party without manual collection, thereby solving the problem of low efficiency and high cost of manually collecting network fingerprint features near the transaction party, and obtaining the network fingerprint features corresponding to the transaction party based on model prediction can improve the accuracy of network fingerprint collection.

[0042] The following detailed description is provided with reference to specific embodiments. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims. The flowcharts shown in the accompanying drawings are for illustrative purposes only and are not necessarily executed in accordance with the steps shown. For example, some steps are arranged in parallel and do not have a strict logical order, so the actual execution order is variable.

[0043] See Figure 1, is a flow chart of a network fingerprint mining model training method provided by an embodiment of the present application. In a specific embodiment, the network fingerprint mining model training method can be applied to a network fingerprint mining model training device or an electronic device equipped with a network fingerprint mining model training device. The following will take the execution subject as an electronic device as an example to illustrate the specific process of this embodiment. Figure 2 The process shown in FIG. 1 is described in detail. The network fingerprint mining model training method may specifically include the following steps:

[0044] S101, obtaining a first sample data set corresponding to a transaction party and a true sample score corresponding to each first sample data point, wherein the first sample data includes a first network data point, a positive sample label or a negative sample label corresponding to the first network data point, and delivery attribute information corresponding to the first network data point, wherein the first network data point includes at least one wireless network;

[0045] It should be noted that when a delivery party executes a delivery task issued by a transaction party, it periodically reports a first network data point at a certain time interval. The first network data point is the nearby network information collected by the delivery party before reporting the first network data point. The first network data point includes at least one wireless network, which can be a Wi-Fi network. For example, after receiving a delivery task issued by transaction party B, delivery party A will rush to the transaction party to obtain the delivery object corresponding to the delivery task. During the journey from delivery party A to transaction party B, the delivery party will periodically collect nearby wireless network information and report at least one wireless network collected each time as a network data point to the server.

[0046] It should be noted that the network fingerprint in the embodiment of the present application can specifically be a Wi-Fi fingerprint.

[0047] See Figure 2 , is a schematic diagram of a scenario for reporting network data points provided by an embodiment of the present application. Figure 2 As shown, after the delivery party receives the delivery task issued by the transaction party, while rushing to the transaction party's location, the delivery party's terminal device scans nearby wireless networks and periodically reports the scanned wireless networks as network data points to the transaction party.

[0048] Delivery attribute information can include the transaction party type corresponding to the delivery task performed by the delivery party, as well as information about certain time points during the overall delivery process. For example, when the transaction party is a merchant, the transaction party type could be: catering merchant, daily necessities merchant, supermarket, convenience store, etc. Time point information can include key time points during the delivery process, such as the time the delivery party arrives at the transaction party and the time the delivery party leaves the transaction party.

[0049] Positive and negative sample labels indicate whether a first network data point can be used as a network fingerprint corresponding to the transaction party, and whether the delivery party can be accurately located near the transaction party based on the first network data point. When a first network data point includes a positive sample label, it indicates that the first network data point can be used as a network fingerprint corresponding to the transaction party, and the delivery party can be accurately located near the transaction party based on the first network data point. When a first network data point includes a negative sample label, it indicates that the first network data point cannot be used as a network fingerprint corresponding to the transaction party, and the delivery party cannot be accurately located near the transaction party.

[0050] It should be noted that when the delivery party executes the delivery task issued by the transaction party, the delivery party will periodically report the first network data point according to a certain time period. The transaction party is equipped with a background device that periodically sends beacon messages. When the delivery party arrives at the transaction party's location, the delivery party's terminal device will receive the beacon message sent by the transaction party's background device. The beacon message is used to indicate that the delivery party has arrived near the transaction party. At this time, the first network data point reported by the delivery party will be set with a positive sample label. When the delivery party has not arrived at the transaction party's location, the delivery party's terminal device will not receive the beacon message pushed by the transaction party's background device. At this time, the first network data point reported by the delivery party will be set with a negative sample label.

[0051] Among them, beacon messages are data messages pushed to devices entering a pre-created signal area based on low-power Bluetooth technology.

[0052] Specifically, the first network data point reported by the delivery party when executing the delivery task issued by the transaction party, the delivery attribute information corresponding to the first network data point, and the positive sample label or negative sample label corresponding to the first network data point are used as the first sample data, and a large amount of first sample data generated by the delivery party when executing the delivery task issued by the transaction party in history is obtained to obtain the first sample data set.

[0053] The true sample score is the sample score corresponding to the first sample data calculated based on the true value data collected offline. Specifically, the network data points corresponding to each location point near the transaction party are collected offline, and the network data points include at least one wireless network and the network signal strength corresponding to the wireless network. The offline collected network data point corresponding to the first network data point is determined based on the wireless network included in the first network data point and the network signal strength corresponding to the wireless network. Then, the location point corresponding to the first sample data is determined based on the correspondence between the first network data point and the offline collected network data point. The score is determined based on the distance between the location point and the transaction party, and the score is used as the true sample score corresponding to the first sample data. For example, after determining the location point corresponding to the first sample data, based on the distance between the location point and the transaction party, if the location point corresponding to the first sample data is 5 meters away from the transaction party, the true sample score corresponding to the first sample data is determined to be 9 points. Generally, the closer the distance between the location point and the transaction party, the higher the true sample score corresponding to the first sample data.

[0054] In one embodiment, obtaining a first sample data set corresponding to the transaction party includes: obtaining a first network data point reported by the delivery party when executing a delivery task issued by the transaction party; determining delivery attribute information corresponding to the first network data point based on the delivery task, the delivery attribute information including the transaction party type and time node information during the delivery process; judging whether the delivery party receives monitoring information sent by the transaction party device when reporting the first network data point, and if the delivery party receives monitoring information sent by the transaction party device when reporting the first network data point, setting a positive sample label for the first network data point; and if the delivery party does not receive monitoring information sent by the transaction party device when reporting the first network data point, setting a negative sample label for the first network data point; generating the first sample data based on the first network data point, the delivery attribute information corresponding to the first network data point, and the positive sample label or negative sample label corresponding to the first network data point.

[0055] S102: training a pre-created first network fingerprint mining model based on the first sample data set and the real sample scores corresponding to each first sample data to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges;

[0056] It should be noted that the network fingerprint mining model is used to predict the network fingerprint features corresponding to the transaction party based on the sample data provided by the transaction party. The network fingerprint features are used to determine whether the network data points reported by the delivery party can be used as the network fingerprint corresponding to the transaction party when the delivery party executes the delivery task issued by the transaction party. When the network data points reported by the delivery party can be used as the network fingerprint corresponding to the transaction party, it is determined that the delivery party has arrived at the location of the transaction party. When the network data points reported by the delivery party cannot be used as the network fingerprint corresponding to the transaction party, it is determined that the delivery party has not arrived at the location of the transaction party.

[0057] Specifically, step S102 inputs a first sample data set corresponding to the transaction party into a pre-created first network fingerprint mining model. The network fingerprint mining model generates a first positive sample data cluster corresponding to the transaction party based on the first sample data set, a first predicted score corresponding to each first sample data point in the first positive sample data cluster, and a first positive sample score threshold determined based on the distribution characteristics of the first predicted scores corresponding to each first sample data point in the first positive sample data cluster. The first positive sample data cluster includes at least one first sample data point, the first predicted score indicates whether the first sample data point is positive, and the first positive sample score threshold is a score threshold used to determine whether the first sample data point is positive. The first sample data point being positive means that the first sample data point can serve as the network fingerprint corresponding to the transaction party, and the delivery party has already arrived at the transaction party's location when reporting the first network data point in the first sample data point. A first difference between the first predicted score and the true sample score is then calculated, and the model parameters of the first network fingerprint mining model are adjusted based on the first difference. Training is iteratively performed until the first difference is less than a first preset threshold, at which point training terminates, resulting in a trained second network fingerprint mining model.

[0058] In one embodiment, the first network fingerprint mining model includes a first clustering module, a first classification prediction module, and a first threshold prediction module. The pre-created first network fingerprint mining model is trained based on the first sample data set and the real sample scores corresponding to each of the first sample data to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges, including: clustering each first sample data in the first sample data set based on the first clustering module to obtain at least one first sample data cluster corresponding to the transaction party; classifying the at least one first sample data cluster based on the first classification prediction module to obtain at least one first positive sample data cluster. data cluster; calculating the first prediction score corresponding to each first sample data in at least one first positive sample data cluster based on the first threshold prediction module, and determining the first positive sample score threshold according to the first prediction score corresponding to each first sample data in the first positive sample data cluster; calculating the first difference between the first prediction score and the true sample score; if the first difference is greater than or equal to the first preset threshold, adjusting the model parameters of the first network fingerprint mining model based on the first difference, and executing the step of clustering each first sample data in the first sample data set; if the first difference is less than the first preset threshold, determining that the first network fingerprint mining model has converged, and obtaining the second network fingerprint mining model.

[0059] S103: Input the first sample data set into a second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster;

[0060] Specifically, after the first network fingerprint mining model is trained to obtain the second network fingerprint mining model, the first sample data set is input into the second network fingerprint mining model to obtain at least one second positive sample data cluster predicted by the second network fingerprint mining model for the first sample data set and the second prediction score corresponding to each first sample data in the second positive sample data cluster.

[0061] In one embodiment, the second network fingerprint mining model includes a second clustering module, a second classification prediction module, and a second threshold prediction module. The first sample data set is input into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster, including: clustering each first sample data in the first sample data set based on the second clustering module to obtain at least one second sample data cluster corresponding to the transaction party; classifying at least one second sample data cluster based on the second classification prediction module to obtain at least one second positive sample data cluster; and calculating the second prediction score corresponding to each first sample data in at least one second positive sample data cluster based on the second threshold prediction module.

[0062] It is worth mentioning that the first sample data set is a sample data set with a high density of positive sample data, in which the positive sample data all contain corresponding positive sample labels. The main reason for using a sample data set with a high density of positive sample data to train the first network fingerprint mining model is that the sample data set with a low density of positive sample data relies too much on expert rules during the model training process, and the trained model often has low accuracy and low coverage. The first network fingerprint mining model is trained based on the first sample data set with a high density of positive sample data to obtain a trained second network fingerprint mining model. The second network fingerprint mining model can obtain positive sample data clusters and positive sample score thresholds for transaction parties with a high density of positive sample data based on the distribution of positive sample data and other characteristics.

[0063] However, in actual offline scenarios, the sample data set corresponding to the transaction party may not have positive sample label data or the positive sample data density is low and the noise is high. At this time, the second network fingerprint mining model based on the training does not have a good model coverage effect for the transaction party with a low positive sample data density. At this time, a third network fingerprint mining model is pre-created, and the initial model parameters of the third network fingerprint mining model are exactly the same as those of the second network fingerprint mining model. The labels in the second positive sample data cluster in the second network fingerprint mining model are removed to obtain a sample data cluster that does not contain positive sample labels and negative sample labels. The sample data cluster that does not contain positive sample labels and negative sample labels is used as input, and the prediction result of the second network fingerprint mining model on the first sample data set is used as the verification result. Finally, the third network fingerprint mining model is trained so that the prediction result of the third network fingerprint mining model for the sample data cluster that does not contain positive sample labels and negative sample labels is close to the second network fingerprint mining model. Since the sample data cluster input by the third network fingerprint mining model does not contain positive sample features and negative sample features, the trained fourth network fingerprint mining model has a better prediction effect for the transaction party with a low positive sample data density. Please refer to the detailed description in S104 for details.

[0064] S104, training the pre-created third network fingerprint mining model based on the second positive sample data cluster and the second prediction scores corresponding to each first sample data in the second positive sample data cluster, to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges, and the model structure and model parameters of the third network fingerprint mining model are the same as those of the second network fingerprint mining model.

[0065] Specifically, after inputting the first sample data set into the second network fingerprint mining model and obtaining at least one second positive sample data cluster corresponding to the first sample data set and the second prediction score corresponding to each first sample data in the second positive sample data cluster, the pre-created third network fingerprint mining model is trained according to the second positive sample data cluster and the second prediction score corresponding to each first sample data in the second positive sample data cluster to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges.

[0066] In one embodiment, the pre-created third network fingerprint mining model is trained based on the second positive sample data cluster and the second prediction scores corresponding to each first sample data in the second positive sample data cluster to obtain a fourth network fingerprint mining model after the third network fingerprint mining model training converges, including: performing label removal processing on each first sample data in the second positive sample data cluster to obtain second sample data that does not contain positive sample labels and negative sample labels corresponding to each first sample data, clustering each second sample data to generate a third sample data cluster corresponding to the second positive sample data cluster; training the pre-created third network fingerprint mining model based on the second prediction scores corresponding to each first sample data in the third sample data cluster and the second positive sample data cluster to obtain a fourth network fingerprint mining model after the third network fingerprint mining model training converges.

[0067] In one embodiment, the third network fingerprint mining model includes a third clustering module, a third classification prediction module, and a third threshold prediction module. The pre-created third network fingerprint mining model is trained based on the second prediction scores corresponding to each first sample data in the third sample data cluster and the second positive sample data cluster to obtain a fourth network fingerprint mining model after the third network fingerprint mining model training converges, including: classifying the third sample data cluster based on the third classification prediction module to obtain at least one third positive sample data cluster; calculating the third prediction scores corresponding to each second sample data in the positive sample data cluster based on the third threshold prediction module, and determining the second positive sample score threshold based on the third prediction scores corresponding to each second sample data in the positive sample data cluster; calculating the second difference between the third prediction score and the second prediction score; if the second difference is greater than or equal to the second preset threshold, adjusting the model parameters of the third network fingerprint mining model based on the second difference, and executing the step of classifying the third sample data cluster; if the second difference is less than the second preset threshold, determining that the third network fingerprint mining model has converged to obtain the fourth network fingerprint mining model.

[0068] In an embodiment of the present application, a first sample data set corresponding to a transaction party and a real sample score corresponding to each first sample data are first obtained, and then a pre-created first network fingerprint mining model is trained based on the first sample data set and the real sample score corresponding to each first sample data to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges. The first sample data set is then input into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster. Finally, a pre-created third network fingerprint mining model is trained based on the second positive sample data cluster and the second prediction score corresponding to each first sample data in the second positive sample data cluster to obtain a fourth network fingerprint mining model after the third network fingerprint mining model training converges. By using the embodiment of the present application, a network fingerprint mining model can be finally obtained that can predict the network fingerprint features corresponding to the transaction party without manual collection, thereby solving the problem of low efficiency and high cost of manually collecting network fingerprint features near the transaction party, and obtaining the network fingerprint features corresponding to the transaction party based on model prediction can improve the accuracy of network fingerprint collection.

[0069] See Figure 3 , is a flow chart of a network fingerprint mining model training method provided in an embodiment of the present application. Figure 3 As shown, the network fingerprint mining model training method may include the following steps S201 to S205:

[0070] S201: Obtain a first sample data set corresponding to a transaction party and a true sample score corresponding to each first sample data point, wherein the first sample data includes a first network data point, a positive sample label or a negative sample label corresponding to the first network data point, and delivery attribute information corresponding to the first network data point, wherein the first network data point includes at least one wireless network;

[0071] Specifically, for step S201, please refer to the detailed description of step S101 in another embodiment of the present application, which will not be repeated here.

[0072] In one embodiment, obtaining a first sample data set corresponding to a transaction party may include the following steps:

[0073] S2011, obtaining the first network data point reported by the delivery party when executing the delivery task issued by the transaction party;

[0074] Specifically, when executing a delivery task issued by the transaction party, the delivery party periodically reports a first network data point at a specific interval. This first network data point is wireless network information surrounding the delivery party collected in real time by the delivery party's terminal device. The first network data point includes at least one wireless network. The transaction party's backend device verifies whether the first network data point matches the transaction party's corresponding network fingerprint, thereby determining whether the delivery party has arrived at the transaction party's location.

[0075] For example, after receiving a delivery task issued by the transaction party, the delivery party rushes from its current location to the transaction party's location to obtain the delivery object corresponding to the delivery task. During this process, the delivery party's terminal device will report the Wi-Fi networks near the delivery party detected by the terminal device to the transaction party's backend device every 5 seconds. Each time, at least one Wi-Fi network reported is sent to the transaction party's backend device as a first network data point. When the delivery party arrives at the transaction party's location, the Wi-Fi network collected by the delivery party's terminal device is the Wi-Fi network near the transaction party. The transaction party can then determine that the delivery party has arrived at the transaction party's location based on the first network data point reported by the delivery party.

[0076] S2012: Determine delivery attribute information corresponding to the first network data point based on the delivery task, where the delivery attribute information includes the transaction party type and time node information during the delivery process;

[0077] Specifically, the delivery party receives the delivery task issued by the transaction party, and can obtain the type information of the transaction party that issued the delivery task based on the delivery task. In the process of executing the delivery task, the delivery service system will record some key time node information, such as the time node when the delivery party arrives at the transaction party's location and actively sends the arrival information to the delivery service system, and the time node when the delivery party actively sends the departure information to the delivery service system after obtaining the delivery object corresponding to the delivery task. Based on the time node of sending the arrival information and the time node of sending the departure information, the waiting time of the delivery party at the transaction party can also be calculated.

[0078] It should be noted that when the delivery party arrives at the location of the transaction party during the execution of the delivery task, the delivery party will send arrival information to the delivery service system through the terminal device on the service interface of the delivery task, and when the delivery party obtains the delivery object corresponding to the delivery task, it will send arrival information to the delivery service system through the terminal device on the service interface of the delivery task.

[0079] For example, in the food delivery service industry, the delivery party is the rider and the transaction party is the food delivery merchant. When the rider receives a delivery task issued by the food delivery merchant, he can obtain the merchant registration information of the food delivery merchant based on the delivery task. The merchant registration information includes merchant type information, and when the rider is performing the delivery task, he first goes to the food delivery merchant to get the food. Then, after the rider arrives at the merchant store, the rider can click on "Arrived at the merchant" through the delivery task-related service interface on the mobile phone. When the rider gets the food and leaves the merchant, he can click on "Left the merchant" through the delivery task-related service interface on the mobile phone to enable the delivery service system to obtain the delivery status of the current delivery task.

[0080] It is worth mentioning that, generally speaking, the closer the time node to which the delivery party sends the arrival information or the departure information, the greater the possibility that the first network data point reported is a positive sample data point.

[0081] S2013, determining whether the delivery party receives monitoring information sent by the transaction party device when reporting the first network data point; if the delivery party receives monitoring information sent by the transaction party device when reporting the first network data point, setting a positive sample label for the first network data point; if the delivery party does not receive monitoring information sent by the transaction party device when reporting the first network data point, setting a negative sample label for the first network data point;

[0082] Transaction party equipment refers to the background equipment set up within the transaction party for managing the transaction party's affairs.

[0083] It should be noted that when the delivery party executes the delivery task issued by the transaction party, the delivery party will periodically report the first network data point according to a certain time period. The transaction party is equipped with a background device that periodically sends detection information. When the delivery party arrives at the transaction party's location, the delivery party's terminal device will receive the detection information sent by the transaction party's background device. The detection information is used to indicate that the delivery party has arrived near the transaction party. At this time, the first network data point reported by the delivery party will be set with a positive sample label. When the delivery party has not arrived at the transaction party's location, the delivery party's terminal device will not receive the detection information pushed by the transaction party's background device. At this time, the first network data point reported by the delivery party will be set with a negative sample label.

[0084] S2014: Generate first sample data based on the first network data point, the delivery attribute information corresponding to the first network data point, and the positive sample label or negative sample label corresponding to the first network data point.

[0085] S202: training a pre-created first network fingerprint mining model based on the first sample data set and the real sample scores corresponding to each first sample data to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges;

[0086] In an embodiment of the present application, a network fingerprint mining model is used to predict the network fingerprint features corresponding to the transaction party based on sample data provided by the transaction party. The network fingerprint features are used to determine whether the network data points reported by the delivery party can be used as the network fingerprint corresponding to the transaction party when the delivery party executes the delivery task issued by the transaction party. When the network data points reported by the delivery party can be used as the network fingerprint corresponding to the transaction party, it is determined that the delivery party has arrived at the location of the transaction party. When the network data points reported by the delivery party cannot be used as the network fingerprint corresponding to the transaction party, it is determined that the delivery party has not arrived at the location of the transaction party.

[0087] See Figure 4 , is a flow chart of a network fingerprint mining model training method provided in an embodiment of the present application. Figure 4 As shown, step S202 may include the following steps:

[0088] S2021: performing clustering processing on each first sample data in the first sample data set based on the first clustering module to obtain at least one first sample data cluster corresponding to the transaction party;

[0089] In the embodiment of the present application, the first network fingerprint mining model includes a first clustering module, a first classification prediction module, and a first threshold prediction module.

[0090] Specifically, step S2021 inputs the first sample data set into a first network fingerprint mining model, and clusters each first sample data in the first sample data set based on a first clustering module in the first network fingerprint mining model to obtain at least one first sample data cluster after clustering. The first sample data cluster includes at least one first sample data.

[0091] It should be noted that, in the embodiment of the present application, the first clustering module adopts a clustering algorithm based on a graph structure, which can use the sparse structure to reflect the distribution of samples and can also provide global attribute features that describe the clustering results.

[0092] S2022, performing classification processing on at least one first sample data cluster based on the first classification prediction module to obtain at least one first positive sample data cluster;

[0093] Specifically, each first sample data cluster obtained after clustering is input into a first classification prediction module. Based on the first classification prediction module and in combination with preset empirical rules, a high-confidence first positive sample data cluster is selected from each first sample data cluster. A first positive sample data cluster is a data cluster with a relatively high proportion of positive sample data. Positive sample data refers to first sample data corresponding to a first network data point that can serve as a network fingerprint corresponding to a transaction party.

[0094] S2023, calculating, based on the first threshold prediction module, first prediction scores corresponding to respective first sample data in at least one first positive sample data cluster, and determining a first positive sample score threshold according to the first prediction scores corresponding to respective first sample data in the first positive sample data cluster;

[0095] Specifically, after obtaining a first positive sample data cluster with high confidence, the first threshold prediction module in the first network fingerprint mining model calculates the first prediction score corresponding to each first sample data in the first positive sample data cluster based on a preset empirical function, and determines the first positive sample score threshold according to the distribution of the positive sample data in the first positive sample data cluster and the first prediction score corresponding to each first sample data.

[0096] It should be noted that the first positive sample score threshold is a score threshold used to determine whether the first sample data can be used as positive sample data. When the first prediction score corresponding to the first sample data is greater than the first positive sample score threshold, the first sample data is determined to be positive sample data.

[0097] In one embodiment, step S2023 may include the following steps:

[0098] S231, calculating the network score corresponding to each network in each first sample data in the first positive sample data cluster;

[0099] Specifically, the network score corresponding to each network in the first sample data is calculated according to a preset score calculation function.

[0100] S232, determining a network weight corresponding to each network based on the network signal strength information and network frequency information corresponding to each network;

[0101] Specifically, the network frequency information is the frequency at which the network appears in the first positive sample data cluster. According to the network signal strength information and network frequency information corresponding to each network, the network weight corresponding to each network can be determined.

[0102] S233: Perform weighted summation on the network scores corresponding to the respective networks based on the network weights to obtain a second prediction score corresponding to the first sample data.

[0103] S2024, calculating a first difference between the first prediction score and the true sample score;

[0104] Specifically, the true sample score is the true score corresponding to the first sample data calculated based on the network information collected offline, the first predicted score is the predicted score corresponding to the first sample data predicted based on the first network fingerprint mining model, the first predicted score and the true sample score are compared, and the first difference between the first predicted score and the true sample score is used as the model loss of the first network mining model.

[0105] S2025, if the first difference is greater than or equal to the first preset threshold, adjusting the model parameters of the first network fingerprint mining model based on the first difference, and performing a step of clustering each first sample data in the first sample data set based on the first clustering module;

[0106] S2026: If the first difference is less than the first preset threshold, it is determined that the first network fingerprint mining model has converged, and a second network fingerprint mining model is obtained.

[0107] S203: Input the first sample data set into a second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster;

[0108] It should be noted that the first sample data set is a sample data set with a higher density of positive sample data, in which the positive sample data all contain corresponding positive sample labels. The main reason for using a sample data set with a higher density of positive sample data to train the first network fingerprint mining model is that the sample data set with a lower density of positive sample data relies too much on expert rules during the model training process, and the trained model often has low accuracy and low coverage. The first network fingerprint mining model is trained based on the first sample data set with a higher density of positive sample data to obtain a trained second network fingerprint mining model. The second network fingerprint mining model can obtain positive sample data clusters and positive sample score thresholds for transaction parties with a higher density of positive sample data based on the distribution of positive sample data and other characteristics.

[0109] However, in actual offline scenarios, the sample data set corresponding to the transaction party may not have positive sample label data or the positive sample data density is low and the noise is high. At this time, the second network fingerprint mining model based on the training does not have a good model coverage effect for the transaction party with a low positive sample data density. At this time, a third network fingerprint mining model is pre-created. The initial model parameters of the third network fingerprint mining model are exactly the same as those of the second network fingerprint mining model. The labels in the second positive sample data cluster in the second network fingerprint mining model are removed to obtain a sample data cluster that does not contain positive sample labels and negative sample labels. The sample data cluster that does not contain positive sample labels and negative sample labels is used as input, and the prediction result of the second network fingerprint mining model on the first sample data set is used as the verification result. Finally, the third network fingerprint mining model is trained so that the prediction result of the third network fingerprint mining model for the sample data cluster that does not contain positive sample labels and negative sample labels is close to the second network fingerprint mining model. Since the sample data cluster input by the third network fingerprint mining model does not contain positive sample features and negative sample features, the trained fourth network fingerprint mining model has a better prediction effect for the transaction party with a low positive sample data density.

[0110] See Figure 5 , is a flow chart of a network fingerprint mining model training method provided in an embodiment of the present application. Figure 5 As shown, step S203 may include the following steps:

[0111] S2031: performing clustering processing on each first sample data in the first sample data set based on a second clustering module to obtain at least one second sample data cluster corresponding to a transaction party;

[0112] In the embodiment of the present application, the second network fingerprint mining model includes a second clustering module, a second classification prediction module, and a second threshold prediction module.

[0113] Specifically, the first sample data set is input into the second network fingerprint mining model that has been trained and converged, and the second clustering module in the second network fingerprint mining model clusters the first sample data set to obtain at least one second sample data cluster corresponding to the transaction party.

[0114] S2032, performing classification processing on at least one second sample data cluster based on a second classification prediction module to obtain at least one second positive sample data cluster;

[0115] The second positive sample data cluster is a first sample data set with a relatively high proportion of positive sample data. Positive sample data refers to first sample data corresponding to the first network data point that can serve as the network fingerprint corresponding to the transaction party.

[0116] S2033: Calculate, based on the second threshold prediction module, second prediction scores corresponding to each first sample data in at least one second positive sample data cluster.

[0117] S204: performing label removal processing on each first sample data in the second positive sample data cluster to obtain second sample data corresponding to each first sample data, which does not contain a positive sample label and a negative sample label, and clustering each second sample data to generate a third sample data cluster corresponding to the second positive sample data cluster;

[0118] Specifically, label removal processing is performed on each first sample data in the second positive sample data cluster to obtain second sample data that does not contain positive sample labels and negative sample labels corresponding to each first sample data, and each second sample data is clustered to generate a third sample data cluster corresponding to the second positive sample data cluster to remove the positive sample features and negative sample features in the second positive sample data cluster.

[0119] S205 , training the pre-created third network fingerprint mining model based on the second prediction scores corresponding to each first sample data in the third sample data cluster and the second positive sample data cluster, to obtain a fourth network fingerprint mining model after the third network fingerprint mining model training converges.

[0120] Specifically, the third sample data cluster is used as the input of the third network fingerprint mining model, and the prediction results of the second network fingerprint mining model on the first sample data set are used as supervision to train the third network fingerprint mining model, so that the trained fourth network fingerprint mining model has a better prediction effect on transaction parties with a lower positive sample data density.

[0121] See Figure 6 , is a flow chart of a network fingerprint mining model training method provided in an embodiment of the present application. Figure 6 As shown, step S205 may include the following steps:

[0122] S2051, performing classification processing on the third sample data cluster based on the third classification prediction module to obtain at least one third positive sample data cluster;

[0123] In the embodiment of the present application, the third network fingerprint mining model includes a third clustering module, a third classification prediction module, and a third threshold prediction module.

[0124] Specifically, the third sample data cluster is input into a third network fingerprint mining model. A classification prediction is performed on the third sample data cluster based on a third classification prediction module within the third network fingerprint mining model to obtain a third high-confidence positive sample data cluster. The third positive sample data cluster is a set of second sample data with a relatively high proportion of positive sample data. Positive sample data refers to the second sample data corresponding to the first network data point that can serve as the network fingerprint corresponding to the transaction party.

[0125] S2052: Calculate the third prediction score corresponding to each second sample data in the positive sample data cluster based on the third threshold prediction module, and determine the second positive sample score threshold according to the third prediction score corresponding to each second sample data in the positive sample data cluster;

[0126] Specifically, after obtaining a third positive sample data cluster with high confidence, the third threshold prediction module in the third network fingerprint mining model calculates the third prediction scores corresponding to each second sample data in the third positive sample data cluster based on a preset empirical function, and determines the second positive sample score threshold according to the distribution of positive sample data in the third positive sample data cluster and the third prediction scores corresponding to each second sample data.

[0127] It should be noted that the second positive sample score threshold is a score threshold used to determine whether the second sample data can be used as positive sample data. When the third prediction score corresponding to the second sample data is greater than the second positive sample score threshold, the second sample data is determined to be positive sample data.

[0128] S2053, calculating a second difference between the third prediction score and the second prediction score;

[0129] Specifically, the third prediction score is the prediction score of the second sample data predicted by the third network fingerprint mining model, the second prediction score is the prediction score of the first sample data predicted by the third network fingerprint mining model, and the second sample data is the sample data obtained after the first sample data is processed by label removal. According to the correspondence between the second sample data and the first sample data, the second difference between the third prediction score and the second prediction score is calculated, and the second difference is used as the model loss of the third network mining model.

[0130] S2054, if the second difference is greater than or equal to the second preset threshold, adjusting the model parameters of the third network fingerprint mining model based on the second difference, and executing the step of classifying the third sample data cluster based on the third classification prediction module;

[0131] S2055: If the second difference is less than the second preset threshold, it is determined that the third network fingerprint mining model has converged, and a fourth network fingerprint mining model is obtained.

[0132] In an embodiment of the present application, a first sample data set corresponding to a transaction party and a real sample score corresponding to each first sample data are first obtained, and then a pre-created first network fingerprint mining model is trained based on the first sample data set and the real sample score corresponding to each first sample data to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges, and then the first sample data set is input into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster, and then a label removal process is performed on each first sample data in the second positive sample data cluster to obtain the first sample data corresponding to each first sample data. The second sample data does not contain positive sample labels and negative sample labels, and each second sample data is clustered to generate a third sample data cluster corresponding to the second positive sample data cluster. Finally, the pre-created third network fingerprint mining model is trained based on the second prediction scores corresponding to each first sample data in the third sample data cluster and the second positive sample data cluster, to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges. By adopting the embodiment of the present application, a network fingerprint mining model that can predict the corresponding network fingerprint features of the transaction party without manual collection can be finally obtained, which solves the problem of low efficiency and high cost of manually collecting network fingerprint features near the transaction party, and the network fingerprint features corresponding to the transaction party obtained based on model prediction can improve the accuracy of network fingerprint collection.

[0133] See Figure 7 , is a flow chart of a network fingerprint mining method provided in an embodiment of the present application, the network fingerprint mining method may include the following steps:

[0134] S301, obtaining a third sample data set corresponding to a target transaction party, the third sample data including a third network data point and delivery attribute information corresponding to the third network data point, the third network data point including at least one wireless network;

[0135] Specifically, the third network data points reported by the delivery party when executing the delivery task issued by the target transaction party, the delivery attribute information corresponding to the third network data points, and the positive sample labels or negative sample labels corresponding to the third network data points are used as the third sample data, and a large amount of third sample data generated by the delivery party when executing the delivery task issued by the target transaction party in history is obtained to obtain a third sample data set.

[0136] S302. Input the third sample data set into the fourth network fingerprint mining model trained by the network fingerprint mining model training method to obtain the network fingerprint feature corresponding to the target transaction party. The network fingerprint feature includes at least one target positive sample data cluster and a target positive sample score threshold. The network fingerprint feature is used to determine whether the target network data point reported by the delivery party when executing the delivery task issued by the target transaction party can be used as the network fingerprint corresponding to the target transaction party.

[0137] Specifically, the third sample data set corresponding to the target transaction party is input into the fourth network fingerprint mining model trained using the above-described embodiment. The first network fingerprint mining model predicts the network fingerprint features corresponding to the target transaction party based on the third sample data set corresponding to the target transaction party. The network fingerprint features include at least one target positive sample data cluster and a target positive sample score threshold. The network fingerprint features are used to determine whether the target network data points reported by the delivery party when executing the delivery task issued by the target transaction party can be used as the network fingerprint of the target transaction party.

[0138] It is understood that after obtaining the target positive sample data cluster and the target positive sample score threshold corresponding to the target transaction party, when the delivery party executes the delivery task issued by the target transaction party, the target network data point reported by the delivery party when executing the delivery task issued by the target transaction party is obtained. The target network data point includes at least one target network. The data point score corresponding to the target network data point is determined based on the target positive sample data cluster. When the data point score is greater than or equal to the target positive sample score threshold, the target network data point is determined to be the network fingerprint corresponding to the target transaction party. When the target network data point is the network fingerprint corresponding to the target transaction party, the target transaction party can determine that the delivery party has arrived at the transaction party's location.

[0139] Optionally, the determining of the data point score corresponding to the target network data point based on the target positive sample data cluster can be as follows: determining the target network score and target network weight corresponding to at least one target network according to the distribution information of at least one target network in the target positive sample data cluster, and performing weighted summation based on the target network score and the target network weight to obtain the data point score corresponding to the target network data point.

[0140] See Figure 8 , is a flow chart of a network fingerprint mining method provided by an embodiment of the present application. Figure 8 As shown, step S302 may include the following steps:

[0141] S3021: performing clustering processing on each third sample data in the third sample data set based on a fourth clustering module to obtain at least one third sample data cluster corresponding to the target transaction party;

[0142] In the embodiment of the present application, the fourth network fingerprint mining model includes a fourth clustering module, a fourth classification prediction module, and a fourth threshold prediction module.

[0143] Specifically, the third sample data set is input into the fourth network fingerprint mining model that has been trained and converged, and the fourth clustering module in the fourth network fingerprint mining model performs clustering processing on the third sample data set to obtain at least one third sample data cluster corresponding to the transaction party.

[0144] S3022, performing classification processing on at least one third sample data cluster based on the fourth classification prediction module to obtain at least one target positive sample data cluster;

[0145] Specifically, each of the third sample data clusters obtained after clustering is input into a fourth classification prediction module. Based on the fourth classification prediction module and in combination with preset empirical rules, a high-confidence target positive sample data cluster is selected from each of the third sample data clusters. A target positive sample data cluster is a set of third sample data with a relatively high proportion of positive sample data. Positive sample data refers to third sample data corresponding to third network data points that can serve as the network fingerprint of the transaction party.

[0146] S3023, calculating the fourth prediction score corresponding to each third sample data in at least one target positive sample data cluster based on the fourth threshold prediction module, and determining the target positive sample score threshold according to the fourth prediction score corresponding to each third sample data in the target positive sample data cluster.

[0147] Specifically, after obtaining a high-confidence target positive sample data cluster, the fourth threshold prediction module in the fourth network fingerprint mining model calculates the fourth prediction score corresponding to each third sample data in the target positive sample data cluster based on a preset empirical function, and determines the target positive sample score threshold according to the distribution of positive sample data in the target positive sample data cluster and the fourth prediction score corresponding to each third sample data.

[0148] It should be noted that the target positive sample score threshold is used to determine whether the target network data point reported by the delivery party can be used as the network fingerprint corresponding to the target transaction party. When the delivery party executes the delivery task issued by the target transaction party, the target network score corresponding to the target network data point reported by the delivery party is greater than the target positive sample score threshold, it is determined that the target network data point can be used as the network fingerprint corresponding to the target transaction party, and the target transaction party can determine that the delivery party has arrived at the location of the target transaction party.

[0149] In an embodiment of the present application, by obtaining a third sample data set corresponding to the target transaction party and inputting the third sample data set into a fourth network fingerprint mining model trained by a network fingerprint mining model training method, the network fingerprint features corresponding to the target transaction party can be obtained. By using the embodiment of the present application, a network fingerprint mining model can be finally obtained that can predict the network fingerprint features corresponding to the transaction party without manual collection, thereby solving the problem of low efficiency and high cost of manually collecting network fingerprint features near the transaction party, and obtaining the network fingerprint features corresponding to the transaction party based on model prediction can improve the accuracy of network fingerprint collection.

[0150] See Figure 9 , is a structural diagram of a network fingerprint mining model training device provided in an embodiment of the present application. Figure 9 As shown, the network fingerprint mining model training device 1 can be implemented as all or part of an electronic device through software, hardware, or a combination of both. According to some embodiments, the network fingerprint mining model training device 1 includes a first sample acquisition module 11, a first model training module 12, a second sample acquisition module 13, and a second model training module 14, specifically including:

[0151] A first sample acquisition module 11 is configured to acquire a first sample data set corresponding to a transaction party and a true sample score corresponding to each first sample data point, wherein the first sample data includes a first network data point, a positive sample label or a negative sample label corresponding to the first network data point, and delivery attribute information corresponding to the first network data point, wherein the first network data point includes at least one wireless network;

[0152] A first model training module 12 is configured to train a pre-created first network fingerprint mining model based on the first sample data set and the real sample scores corresponding to each of the first sample data, to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges;

[0153] A second sample acquisition module 13 is configured to input the first sample data set into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster;

[0154] The second model training module 14 is used to train the pre-created third network fingerprint mining model based on the second positive sample data cluster and the second prediction scores corresponding to each of the first sample data in the second positive sample data cluster, to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges, and the model structure and model parameters of the third network fingerprint mining model and the second network fingerprint mining model are the same.

[0155] Optionally, when executing the step of acquiring the first sample data set corresponding to the transaction party, the first sample acquisition module 11 is specifically configured to:

[0156] Obtaining a first network data point reported by the delivery party when executing the delivery task issued by the transaction party;

[0157] Determining delivery attribute information corresponding to the first network data point based on the delivery task, the delivery attribute information including transaction party type and time node information during the delivery process;

[0158] Determine whether the delivery party receives monitoring information sent by the transaction party device when reporting the first network data point; if the delivery party receives the monitoring information sent by the transaction party device when reporting the first network data point, set a positive sample label for the first network data point; if the delivery party does not receive the monitoring information sent by the transaction party device when reporting the first network data point, set a negative sample label for the first network data point;

[0159] The first sample data is generated based on the first network data point, the delivery attribute information corresponding to the first network data point, and the positive sample label or negative sample label corresponding to the first network data point.

[0160] Optionally, the first network fingerprint mining model includes a first clustering module, a first classification prediction module, and a first threshold prediction module. The first model training module 12 is specifically configured to:

[0161] performing clustering processing on each of the first sample data in the first sample data set based on the first clustering module to obtain at least one first sample data cluster corresponding to the transaction party;

[0162] Performing classification processing on the at least one first sample data cluster based on the first classification prediction module to obtain at least one first positive sample data cluster;

[0163] Calculating first prediction scores corresponding to respective first sample data in the at least one first positive sample data cluster based on the first threshold prediction module, and determining a first positive sample score threshold according to the first prediction scores corresponding to respective first sample data in the first positive sample data cluster;

[0164] Calculating a first difference between the first prediction score and the true sample score;

[0165] If the first difference is greater than or equal to a first preset threshold, adjusting the model parameters of the first network fingerprint mining model based on the first difference, and performing the step of clustering each of the first sample data in the first sample data set based on the first clustering module;

[0166] If the first difference is less than a first preset threshold, it is determined that the first network fingerprint mining model converges, and a second network fingerprint mining model is obtained.

[0167] Optionally, when the first model training module 12 calculates the first prediction score corresponding to each first sample data in the at least one first positive sample data cluster based on the first threshold prediction module, it is specifically configured to:

[0168] Calculating the network score corresponding to each network in each first sample data in the first positive sample data cluster;

[0169] Determining the network weights corresponding to the networks based on the network signal strength information and the network frequency information corresponding to the networks;

[0170] Based on the network weights, a weighted sum is performed on the network scores corresponding to the networks to obtain a second prediction score corresponding to the first sample data.

[0171] Optionally, the second network fingerprint mining model includes a second clustering module, a second classification prediction module, and a second threshold prediction module. The second sample acquisition module 13 is specifically configured to:

[0172] performing clustering processing on each of the first sample data in the first sample data set based on the second clustering module to obtain at least one second sample data cluster corresponding to the transaction party;

[0173] Performing classification processing on the at least one second sample data cluster based on the second classification prediction module to obtain at least one second positive sample data cluster;

[0174] The second prediction score corresponding to each first sample data in the at least one second positive sample data cluster is calculated based on the second threshold prediction module.

[0175] Optionally, the second model training module 14 is specifically configured to:

[0176] Performing label removal processing on each first sample data in the second positive sample data cluster to obtain second sample data corresponding to each first sample data, which does not contain a positive sample label and a negative sample label, and clustering each second sample data to generate a third sample data cluster corresponding to the second positive sample data cluster;

[0177] The pre-created third network fingerprint mining model is trained based on the second prediction scores corresponding to each of the first sample data in the third sample data cluster and the second positive sample data cluster, to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges.

[0178] Optionally, when the second model training module 14 executes the training of the pre-created third network fingerprint mining model based on the second prediction scores corresponding to each of the first sample data in the third sample data cluster and the second positive sample data cluster to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges, it is specifically used to:

[0179] Performing classification processing on the third sample data cluster based on the third classification prediction module to obtain at least one third positive sample data cluster;

[0180] Calculating, based on the third threshold prediction module, third prediction scores corresponding to each second sample data in the positive sample data cluster, and determining a second positive sample score threshold according to the third prediction scores corresponding to each second sample data in the positive sample data cluster;

[0181] calculating a second difference between the third prediction score and the second prediction score;

[0182] If the second difference is greater than or equal to a second preset threshold, adjusting the model parameters of the third network fingerprint mining model based on the second difference, and performing the step of classifying the third sample data cluster based on the third classification prediction module;

[0183] If the second difference is less than a second preset threshold, it is determined that the third network fingerprint mining model has converged, and a fourth network fingerprint mining model is obtained.

[0184] The order reporting device provided by the embodiment of the present application is used to first obtain a first sample data set corresponding to the transaction party and a real sample score corresponding to each first sample data. Then, based on the first sample data set and the real sample score corresponding to each first sample data, a pre-created first network fingerprint mining model is trained to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges. Then, the first sample data set is input into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster. Then, label removal processing is performed on each first sample data in the second positive sample data cluster to obtain each first sample data. According to the corresponding second sample data that does not contain positive sample labels and negative sample labels, each second sample data is clustered to generate a third sample data cluster corresponding to the second positive sample data cluster, and finally, based on the second prediction scores corresponding to each first sample data in the third sample data cluster and the second positive sample data cluster, a pre-created third network fingerprint mining model is trained to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges; by using the embodiment of the present application, a network fingerprint mining model that can predict the corresponding network fingerprint features of the transaction party without manual collection can be finally obtained, thereby solving the problem of low efficiency and high cost of manually collecting network fingerprint features near the transaction party, and obtaining the network fingerprint features corresponding to the transaction party based on model prediction can improve the accuracy of network fingerprint collection.

[0185] See Figure 10 , is a schematic diagram of the structure of a network fingerprint mining device provided in an embodiment of the present application. Figure 10 As shown, the network fingerprint mining device 1 can be implemented as all or part of an electronic device through software, hardware, or a combination of both. According to some embodiments, the network fingerprint mining device 1 includes a third sample acquisition module 21 and a fingerprint feature determination module 22, specifically including:

[0186] A third sample acquisition module 21 is configured to acquire a third sample data set corresponding to a target transaction party, wherein the third sample data includes a third network data point and delivery attribute information corresponding to the third network data point, wherein the third network data point includes at least one wireless network;

[0187] The fingerprint feature determination module 22 is used to input the third sample data set into the fourth network fingerprint mining model that has been trained using the network fingerprint mining model training method according to any one of claims 1 to 6, to obtain the network fingerprint feature corresponding to the target transaction party, wherein the network fingerprint feature includes at least one target positive sample data cluster and a target positive sample score threshold, and the network fingerprint feature is used to determine whether the target network data point reported by the delivery party when executing the delivery task issued by the target transaction party can be used as the network fingerprint corresponding to the target transaction party.

[0188] Optionally, the fingerprint feature determination module 22 is specifically configured to:

[0189] performing clustering processing on each of the third sample data in the third sample data set based on the fourth clustering module to obtain at least one third sample data cluster corresponding to the target transaction party;

[0190] Performing classification processing on the at least one third sample data cluster based on the fourth classification prediction module to obtain at least one target positive sample data cluster;

[0191] Based on the fourth threshold prediction module, the fourth prediction scores corresponding to each third sample data in the at least one target positive sample data cluster are calculated, and the target positive sample score threshold is determined according to the fourth prediction scores corresponding to each third sample data in the target positive sample data cluster.

[0192] In an embodiment of the present application, by obtaining a third sample data set corresponding to the target transaction party and inputting the third sample data set into a fourth network fingerprint mining model trained by a network fingerprint mining model training method, the network fingerprint features corresponding to the target transaction party can be obtained. By using the embodiment of the present application, a network fingerprint mining model can be finally obtained that can predict the network fingerprint features corresponding to the transaction party without manual collection, thereby solving the problem of low efficiency and high cost of manually collecting network fingerprint features near the transaction party, and obtaining the network fingerprint features corresponding to the transaction party based on model prediction can improve the accuracy of network fingerprint collection.

[0193] The present application also provides a computer storage medium, which can store multiple instructions, and the instructions are suitable for being loaded and executed by a processor as described above. Figures 1 to 8 The network fingerprint mining model training method of the embodiment shown in the figure can be found in the specific execution process. Figures 1 to 8 The detailed description of the illustrated embodiment will not be repeated here.

[0194] The present application also provides a computer program product, which stores at least one instruction, and the at least one instruction is loaded and executed by the processor as described above. Figures 1 to 8 The network fingerprint mining model training method of the embodiment shown in the figure can be found in the specific execution process. Figures 1 to 8 The detailed description of the illustrated embodiment will not be repeated here.

[0195] Please refer to Figure 11 , shows a block diagram of the structure of an electronic device provided by an exemplary embodiment of the present application. The electronic device in the present application may include one or more of the following components: a processor 110, a memory 120, an input device 130, an output device 140, and a bus 150. The processor 110, the memory 120, the input device 130, and the output device 140 may be connected via the bus 150.

[0196] The processor 110 may include one or more processing cores. The processor 110 utilizes various interfaces and circuits to connect various components within the terminal. It executes instructions, programs, code sets, or instruction sets stored in the memory 120, as well as accesses data stored in the memory 120, to perform various functions of the terminal 100 and process data. Optionally, the processor 110 may be implemented using at least one of the following hardware forms: a digital signal processing (DSP), a field-programmable gate array (FPGA), or a programmable logic array (PLA). The processor 110 may integrate one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. The CPU primarily processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing display content; and the modem handles wireless communications. It is understood that the modem may not be integrated into the processor 110 and may be implemented separately via a communications chip.

[0197] The memory 120 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 120 includes a non-transitory computer-readable storage medium. The memory 120 may be used to store instructions, programs, codes, code sets, or instruction sets.

[0198] The input device 130 is used to receive input commands or data and includes, but is not limited to, a keyboard, mouse, camera, microphone, or touch-screen device. The output device 140 is used to output commands or data and includes, but is not limited to, a display device and a speaker. In this embodiment of the present application, the input device 130 may be a temperature sensor for obtaining the operating temperature of the terminal. The output device 140 may be a speaker for outputting audio signals.

[0199] In addition, those skilled in the art will understand that the structure of the terminal shown in the above figures does not constitute a limitation of the terminal. The terminal may include more or fewer components than shown, or a combination of certain components, or a different arrangement of components. For example, the terminal also includes a radio frequency circuit, an input unit, a sensor, an audio circuit, a wireless fidelity (Wi-Fi) module, a power supply, a Bluetooth module, and other components, which will not be described in detail here.

[0200] In the embodiments of the present application, the execution subject of each step can be the terminal described above. Optionally, the execution subject of each step is the operating system of the terminal. The operating system can be Android, iOS, or other operating systems, which are not limited in the embodiments of the present application.

[0201] exist Figure 11 In the electronic device, the processor 110 can be used to call the network fingerprint mining model training program stored in the memory 120 and execute it to implement the network fingerprint mining model training method as described in the various method embodiments of the present application.

[0202] In an embodiment of the present application, a first sample data set corresponding to a transaction party and a real sample score corresponding to each first sample data are first obtained, and then a pre-created first network fingerprint mining model is trained based on the first sample data set and the real sample score corresponding to each first sample data to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges, and then the first sample data set is input into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster, and then a label removal process is performed on each first sample data in the second positive sample data cluster to obtain the first sample data corresponding to each first sample data. The second sample data does not contain positive sample labels and negative sample labels, and each second sample data is clustered to generate a third sample data cluster corresponding to the second positive sample data cluster. Finally, the pre-created third network fingerprint mining model is trained based on the second prediction scores corresponding to each first sample data in the third sample data cluster and the second positive sample data cluster, to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges. By adopting the embodiment of the present application, a network fingerprint mining model that can predict the corresponding network fingerprint features of the transaction party without manual collection can be finally obtained, which solves the problem of low efficiency and high cost of manually collecting network fingerprint features near the transaction party, and the network fingerprint features corresponding to the transaction party obtained based on model prediction can improve the accuracy of network fingerprint collection.

[0203] Those skilled in the art will clearly understand that the technical solution of this application can be implemented with the help of software and / or hardware. "Unit" and "module" in this application refer to software and / or hardware that can independently complete or cooperate with other components to complete specific functions, where the hardware can be, for example, a field programmable gate array (FPGA), an integrated circuit (IC), etc.

[0204] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

[0205] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0206] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of the device or unit can be electrical or other forms.

[0207] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0208] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0209] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable memory, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0210] The above is only an exemplary embodiment of the present application and the scope of the present application cannot be limited thereto. That is, any equivalent changes and modifications made according to the teachings of this application are still within the scope of this application. After considering the disclosure of the specification and practicing here, those skilled in the art will easily think of other embodiments of the present application. This application is intended to cover any variation, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary technical means in the art that are not recorded in this application. The description and examples are to be regarded as exemplary only, and the scope and spirit of this application are defined by the claims.

Claims

1. A network fingerprint mining model training method, characterized in that: The method comprises: Obtaining a first sample data set corresponding to a transaction party and a true sample score corresponding to each of the first sample data, wherein the first sample data includes a first network data point, a positive sample label or a negative sample label corresponding to the first network data point, and delivery attribute information corresponding to the first network data point, wherein the first network data point includes at least one network; Training a pre-created first network fingerprint mining model based on the first sample data set and the real sample scores corresponding to each of the first sample data to obtain a second network fingerprint mining model after the training of the first network fingerprint mining model converges; Inputting the first sample data set into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster; The pre-created third network fingerprint mining model is trained based on the second positive sample data cluster and the second prediction scores corresponding to each of the first sample data in the second positive sample data cluster to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges. The model structure and model parameters of the third network fingerprint mining model and the second network fingerprint mining model are the same.

2. The method according to claim 1, characterized in that The obtaining of a first sample data set corresponding to the transaction party includes: Obtaining a first network data point reported by the delivery party when executing the delivery task issued by the transaction party; Determining delivery attribute information corresponding to the first network data point based on the delivery task, the delivery attribute information including transaction party type and time node information during the delivery process; Determine whether the delivery party receives monitoring information sent by the transaction party device when reporting the first network data point; if the delivery party receives the monitoring information sent by the transaction party device when reporting the first network data point, set a positive sample label for the first network data point; if the delivery party does not receive the monitoring information sent by the transaction party device when reporting the first network data point, set a negative sample label for the first network data point; The first sample data is generated based on the first network data point, the delivery attribute information corresponding to the first network data point, and the positive sample label or negative sample label corresponding to the first network data point.

3. The method according to claim 1, characterized in that The first network fingerprint mining model includes a first clustering module, a first classification prediction module, and a first threshold prediction module. The pre-created first network fingerprint mining model is trained based on the first sample data set and the real sample scores corresponding to each of the first sample data to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges, including: performing clustering processing on each of the first sample data in the first sample data set based on the first clustering module to obtain at least one first sample data cluster corresponding to the transaction party; Performing classification processing on the at least one first sample data cluster based on the first classification prediction module to obtain at least one first positive sample data cluster; Calculating first prediction scores corresponding to respective first sample data in the at least one first positive sample data cluster based on the first threshold prediction module, and determining a first positive sample score threshold according to the first prediction scores corresponding to respective first sample data in the first positive sample data cluster; Calculating a first difference between the first prediction score and the true sample score; If the first difference is greater than or equal to a first preset threshold, adjusting the model parameters of the first network fingerprint mining model based on the first difference, and performing the step of clustering each of the first sample data in the first sample data set based on the first clustering module; If the first difference is less than a first preset threshold, it is determined that the first network fingerprint mining model converges, and a second network fingerprint mining model is obtained.

4. The method according to claim 3, characterized in that The calculating, based on the first threshold prediction module, first prediction scores corresponding to respective first sample data in the at least one first positive sample data cluster includes: Calculating the network score corresponding to each network in each first sample data in the first positive sample data cluster; Determining the network weights corresponding to the networks based on the network signal strength information and the network frequency information corresponding to the networks; Based on the network weights, a weighted sum is performed on the network scores corresponding to the networks to obtain a second prediction score corresponding to the first sample data.

5. The method according to claim 1, wherein The second network fingerprint mining model includes a second clustering module, a second classification prediction module, and a second threshold prediction module. Inputting the first sample data set into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster includes: performing clustering processing on each of the first sample data in the first sample data set based on the second clustering module to obtain at least one second sample data cluster corresponding to the transaction party; Performing classification processing on the at least one second sample data cluster based on the second classification prediction module to obtain at least one second positive sample data cluster; The second prediction score corresponding to each first sample data in the at least one second positive sample data cluster is calculated based on the second threshold prediction module.

6. The method according to claim 1, characterized in that The method of training a pre-created third network fingerprint mining model based on the second positive sample data cluster and the second prediction scores respectively corresponding to each of the first sample data in the second positive sample data cluster to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges includes: Performing label removal processing on each first sample data in the second positive sample data cluster to obtain second sample data corresponding to each first sample data, which does not contain a positive sample label and a negative sample label, and clustering each second sample data to generate a third sample data cluster corresponding to the second positive sample data cluster; The pre-created third network fingerprint mining model is trained based on the second prediction scores corresponding to each of the first sample data in the third sample data cluster and the second positive sample data cluster, to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges.

7. The method according to claim 6, characterized in that The third network fingerprint mining model includes a third clustering module, a third classification prediction module, and a third threshold prediction module. The pre-created third network fingerprint mining model is trained based on the second prediction scores corresponding to each of the first sample data in the third sample data cluster and the second positive sample data cluster, to obtain a fourth network fingerprint mining model after the third network fingerprint mining model training converges, including: Performing classification processing on the third sample data cluster based on the third classification prediction module to obtain at least one third positive sample data cluster; Calculating, based on the third threshold prediction module, third prediction scores corresponding to each second sample data in the positive sample data cluster, and determining a second positive sample score threshold according to the third prediction scores corresponding to each second sample data in the positive sample data cluster; calculating a second difference between the third prediction score and the second prediction score; If the second difference is greater than or equal to a second preset threshold, adjusting the model parameters of the third network fingerprint mining model based on the second difference, and performing the step of classifying the third sample data cluster based on the third classification prediction module; If the second difference is less than a second preset threshold, it is determined that the third network fingerprint mining model has converged, and a fourth network fingerprint mining model is obtained.

8. A network fingerprint mining method, characterized in that: The method comprises: Acquire a third sample data set corresponding to the target transaction party, the third sample data including a third network data point and delivery attribute information corresponding to the third network data point, the third network data point including at least one network; The third sample data set is input into the fourth network fingerprint mining model that has been trained using the network fingerprint mining model training method according to any one of claims 1 to 6 to obtain the network fingerprint feature corresponding to the target transaction party, wherein the network fingerprint feature includes at least one target positive sample data cluster and a target positive sample score threshold, and the network fingerprint feature is used to determine whether the target network data point reported by the delivery party when executing the delivery task issued by the target transaction party can be used as the network fingerprint corresponding to the target transaction party.

9. The method according to claim 8, characterized in that The fourth network fingerprint mining model includes a fourth clustering module, a fourth classification prediction module, and a fourth threshold prediction module. Inputting the third sample data set into the fourth network fingerprint mining model trained by the network fingerprint mining model training method according to any one of claims 1 to 6 to obtain at least one target positive sample data cluster and a target positive sample score threshold corresponding to the target transaction party includes: performing clustering processing on each of the third sample data in the third sample data set based on the fourth clustering module to obtain at least one third sample data cluster corresponding to the target transaction party; Performing classification processing on the at least one third sample data cluster based on the fourth classification prediction module to obtain at least one target positive sample data cluster; Based on the fourth threshold prediction module, the fourth prediction scores corresponding to each third sample data in the at least one target positive sample data cluster are calculated, and the target positive sample score threshold is determined according to the fourth prediction scores corresponding to each third sample data in the target positive sample data cluster.

10. A network fingerprint mining model training device, characterized in that: The device comprises: A first sample acquisition module is configured to acquire a first sample data set corresponding to a transaction party and a true sample score corresponding to each first sample data point, wherein the first sample data includes a first network data point, a positive sample label or a negative sample label corresponding to the first network data point, and delivery attribute information corresponding to the first network data point, wherein the first network data point includes at least one wireless network; A first model training module is configured to train a pre-created first network fingerprint mining model based on the first sample data set and the real sample scores corresponding to each of the first sample data, to obtain a second network fingerprint mining model after the first network fingerprint mining model training converges; A second sample acquisition module is configured to input the first sample data set into the second network fingerprint mining model to obtain at least one second positive sample data cluster corresponding to the first sample data set and a second prediction score corresponding to each first sample data in the second positive sample data cluster; The second model training module is used to train the pre-created third network fingerprint mining model based on the second positive sample data cluster and the second prediction scores corresponding to each of the first sample data in the second positive sample data cluster, to obtain a fourth network fingerprint mining model after the training of the third network fingerprint mining model converges, and the model structure and model parameters of the third network fingerprint mining model and the second network fingerprint mining model are the same.

11. A network fingerprint mining device, characterized in that: The device comprises: A third sample acquisition module is configured to acquire a third sample data set corresponding to the target transaction party, wherein the third sample data includes a third network data point and delivery attribute information corresponding to the third network data point, wherein the third network data point includes at least one wireless network; A fingerprint feature determination module is used to input the third sample data set into a fourth network fingerprint mining model that has been trained using the network fingerprint mining model training method according to any one of claims 1 to 6, to obtain the network fingerprint feature corresponding to the target transaction party, wherein the network fingerprint feature includes at least one target positive sample data cluster and a target positive sample score threshold, and the network fingerprint feature is used to determine whether the target network data point reported by the delivery party when executing the delivery task issued by the target transaction party can be used as the network fingerprint corresponding to the target transaction party.

12. A storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

13. An electronic device comprising: A processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the steps of the method according to any one of claims 1 to 7 or 8 to 9.

Images