CN118075873B

CN118075873B - Positioning method and data processing method based on wireless network data

Info

Publication number: CN118075873B
Application number: CN202410476557.6A
Authority: CN
Inventors: 李向阳; 周颢; 郭楷文; 张延�; 沈国斌
Original assignee: Shanghai Hummingbird Instant Information Technology Co ltd; Zhejiang Koubei Network Technology Co Ltd
Current assignee: Shanghai Hummingbird Instant Information Technology Co ltd; Zhejiang Koubei Network Technology Co Ltd
Priority date: 2024-04-19
Filing date: 2024-04-19
Publication date: 2024-06-21
Anticipated expiration: 2044-04-19
Also published as: CN118075873A

Abstract

The application provides a positioning method and a data processing method based on wireless network data, wherein in the positioning method based on wireless network data, a plurality of scanned list samples comprising wireless network samples provided by a distribution terminal are obtained; obtaining each position label sample corresponding to the scanning time of each scanning list sample of the distribution capacity record provided by the distribution terminal; then, obtaining a scan list feature vector for vector representation of feature information of the scan list sample; and then, obtaining the probability that the scanning time of the scanning list sample is in the second time period, further enabling the scanning list sample to correspond to the position label sample to which the scanning list sample truly belongs, and further enabling the target wireless network positioning model obtained by training the initial wireless network positioning model based on the scanning list feature vector, the probability and the position label sample to obtain more accurate position information and be wider in applicability.

Description

Positioning method and data processing method based on wireless network data

Technical Field

The application relates to the technical field of computers, in particular to a positioning method and a data processing method based on wireless network data.

Background

With the continuous improvement of the quality of life, more and more people are enthusiastically ordering goods on line. And so it is increasingly important to distribute items ordered on the subscriber line.

In the delivery process, positioning the position of the delivery rider becomes an important ring for obtaining an order delivery state, and when the delivery rider is outdoors, the delivery rider can be positioned based on a GPS (Global Positioning System ) of a delivery terminal held by the delivery rider; however, the distribution rider cannot be located based on GPS when the distribution rider is in an indoor environment, such as when the distribution rider is in a mall where a merchant is located. In this case, the delivery rider is typically positioned based on the delivery terminal reporting data to and from the merchant location. However, in the method for positioning the delivery riders, the delivery riders can be positioned only when the delivery riders report that the delivery riders arrive at the merchant positions and leave the merchant positions, and the delivery riders cannot be positioned in continuous time; meanwhile, in the mode of positioning by adopting the data of the report position of the delivery rider, the delivery rider can report to the position of the commercial tenant or leave the position of the commercial tenant in advance, so that the finally obtained positioning information is inaccurate. Therefore, how to accurately locate the distribution rider in the indoor scene is a technical problem to be solved.

Disclosure of Invention

The application provides a positioning method based on wireless network data, which is used for accurately positioning delivery riders in indoor scenes.

The application provides a positioning method based on wireless network data, which is applied to a server, and comprises the following steps: obtaining a plurality of scanned list samples provided by a distribution terminal and containing wireless network samples; wherein each scan list sample comprises a plurality of wireless network samples; obtaining a scanning list feature vector for vector representation of feature information of the scanning list sample according to the hypergraph constructed by the wireless network sample and the scanning list sample; obtaining each position label sample which corresponds to the scanning time of each scanning list sample and is used for representing that the delivery capacity is positioned at a goods taking position and is provided by a delivery terminal, wherein the position label sample comprises first time period information of the delivery capacity record positioned at the goods taking position; the scanning time of the scanning list sample is in a first time period; obtaining the probability that the scanning time of the scanning list sample is in a second time period; the second time period is an actual time period in which the delivery capacity determined based on the first time period is located at the pick-up location; training an initial wireless network positioning model based on the characteristic vector of the scanning list, the probability and the position label sample to obtain a target wireless network positioning model, wherein the target wireless network positioning model is used for obtaining a position label of a position to be positioned according to a scanning list which is provided by a distribution terminal and contains a wireless network and scanned at the position to be positioned.

Optionally, the obtaining a scan list feature vector for vector representation of feature information of the scan list sample according to the hypergraph constructed by the wireless network sample and the scan list sample includes: obtaining a wireless network feature vector for vector representation of feature information of the wireless network sample according to the hypergraph constructed by the wireless network sample and the scanning list sample; and performing attention mechanism calculation on the wireless network feature vector to obtain a scan list feature vector for vector representation of the feature information of the scan list sample.

Optionally, the obtaining a wireless network feature vector for vector representing feature information of the wireless network sample according to the hypergraph constructed by the wireless network sample and the scan list sample includes: constructing a hypergraph representing the association relationship between the wireless network sample and the scanning list sample according to the wireless network sample and the scanning list sample, wherein vertexes in the hypergraph represent the wireless network sample, and superedges in the hypergraph represent the scanning list sample; obtaining a matrix to be encoded containing topological relation information used for representing the wireless network samples in the hypergraph according to the hypergraph; and obtaining a wireless network characteristic vector for carrying out vector representation on the characteristic information of the wireless network sample according to the matrix to be encoded.

Optionally, the method further comprises: obtaining an adjacency matrix for representing the subordinate relation between the wireless network sample and the scanning list sample in the hypergraph according to the network signal intensity information of the hypergraph and the wireless network sample; the obtaining, according to the hypergraph, a matrix to be encoded including information representing a topological relation between the wireless network samples in the hypergraph, including: based on the adjacency matrix, a first degree matrix of the wireless network samples in the hypergraph, and a second degree matrix of the scan list samples in the hypergraph, a matrix to be encoded is obtained that includes information representing a topological relationship between the wireless network samples in the hypergraph.

Optionally, the calculating the attention mechanism of the wireless network feature vector to obtain a scan list feature vector for vector representation of feature information of the scan list sample includes: encoding the wireless network feature vector by adopting a first attention mechanism to obtain an encoded wireless network feature vector; and aggregating the encoded wireless network feature vectors by adopting a second attention mechanism to obtain a scanning list feature vector for vector representation of the feature information of the scanning list sample.

Optionally, the encoding the wireless network feature vector using a first attention mechanism to obtain an encoded wireless network feature vector includes: and encoding the association relation characteristic among the wireless network samples in the scanning list samples into the wireless network characteristic vector by adopting a preset encoder to obtain an encoded wireless network characteristic vector.

Optionally, the aggregating the encoded wireless network feature vectors by using a second attention mechanism to obtain a scan list feature vector for vector representation of feature information of the scan list sample, including: weighting the coded wireless network feature vector by taking the network signal intensity information of the wireless network sample as a weight to obtain a weighted wireless network feature vector; and aggregating the weighted wireless network feature vectors to obtain a scanning list feature vector for vector representation of the feature information of the scanning list sample.

Optionally, the aggregating the weighted wireless network feature vectors to obtain a scan list feature vector for vector representation of feature information of the scan list sample includes: determining a scan list sample of any two adjacent scan times of the plurality of scan list samples; and aggregating the weighted wireless network feature vectors based on the scan list samples of the two adjacent scan times to obtain a scan list feature vector for vector representation of the feature information of the scan list samples.

Optionally, the obtaining the probability that the scanning time of the scanning list sample is within the second period of time includes: taking the scanning list sample as input data of a scanning list probability acquisition model to acquire the probability that the scanning time of the scanning list sample is in a second time period; the scan list probability obtaining model is used for obtaining the probability that the scan time of the scan list is in the actual time period that the delivery capacity is located at the pick-up position corresponding to the scan list according to the scan list.

Optionally, the first time period information includes first time information of arrival at the pick-up location recorded by delivery capacity, second time information of departure from the pick-up location recorded by delivery capacity; the method further comprises the steps of: according to the first time information and the second time information, third time information that the delivery capacity actually reaches the goods taking position and fourth time information that the delivery capacity actually leaves the goods taking position are obtained; and determining the second time period according to the third time information and the fourth time information.

Optionally, the method further comprises: acquiring first time error information of delivery capacity record reaching a goods taking position and second time error information of delivery capacity record leaving the goods taking position; the obtaining, according to the first time information and the second time information, third time information that the delivery capacity actually reaches the pick-up location and fourth time information that the delivery capacity actually leaves the pick-up location includes: obtaining the third time information according to the first time information and the first time error information; and obtaining the fourth time information according to the second time information and the second time error information.

Optionally, the method further comprises: obtaining a target scanning list which is provided by the distribution terminal and scanned at a target to-be-positioned position and contains a target wireless network; and taking the target scanning list as input data of the target wireless network positioning model to obtain a target position label of the target position to be positioned.

The application provides a data processing method, which is applied to a distribution terminal, and comprises the following steps: obtaining a first request message sent by a server side and used for requesting to obtain a target scanning list which is scanned at a target to-be-positioned position and contains a target wireless network; responding to the first request message, and sending the target scanning list to the server; the server is used for taking the target scanning list as input data of a target wireless network positioning model to obtain a target position label of the target position to be positioned; the target wireless network positioning model is obtained for the server by adopting the following modes: obtaining a plurality of scanned list samples provided by a distribution terminal and containing wireless network samples; obtaining a scanning list feature vector for vector representation of feature information of the scanning list sample according to the hypergraph constructed by the wireless network sample and the scanning list sample; obtaining each position label sample which corresponds to the scanning time of each scanning list sample and is used for representing that the delivery capacity is positioned at a goods taking position and is provided by a delivery terminal, wherein the position label sample comprises first time period information of the delivery capacity record positioned at the goods taking position; the scanning time of the scanning list sample is in a first time period; obtaining the probability that the scanning time of the scanning list sample is in a second time period; the second time period is an actual time period in which the delivery capacity determined based on the first time period is located at the pick-up location; training an initial wireless network positioning model based on the characteristic vector of the scanning list, the probability and the position label sample to obtain a target wireless network positioning model, wherein the target wireless network positioning model is used for obtaining a position label of a position to be positioned according to a scanning list which is provided by a distribution terminal and contains a wireless network and scanned at the position to be positioned.

The application provides a positioning device based on wireless network data, which is applied to a server, and comprises: a scan list sample obtaining unit, configured to obtain a plurality of scanned scan list samples including wireless network samples provided by the distribution terminal; wherein each scan list sample comprises a plurality of wireless network samples; a scan list feature vector obtaining unit, configured to obtain a scan list feature vector for vector representation of feature information of the scan list sample according to a hypergraph constructed by the wireless network sample and the scan list sample; the position label sample obtaining unit is used for obtaining each position label sample which corresponds to the scanning time of each scanning list sample and is used for indicating that the delivery capacity is positioned at the goods taking position and is used for obtaining the delivery capacity record provided by the delivery terminal, wherein the position label sample comprises first time period information of the delivery capacity record positioned at the goods taking position; the scanning time of the scanning list sample is in a first time period; a probability obtaining unit, configured to obtain a probability that a scanning time of the scan list sample is within a second time period; the second time period is an actual time period in which the delivery capacity determined based on the first time period is located at the pick-up location; the target wireless network positioning model obtaining unit is used for training an initial wireless network positioning model based on the characteristic vector of the scanning list, the probability and the position label sample to obtain a target wireless network positioning model, wherein the target wireless network positioning model is used for obtaining a position label of a position to be positioned according to a scanning list which is provided by a distribution terminal and contains a wireless network and scanned at the position to be positioned.

The application provides a data processing device, which is applied to a distribution terminal, and comprises: a first request message obtaining unit, configured to obtain a first request message sent by a server and used for requesting to obtain a target scan list including a target wireless network scanned at a target to-be-located position; a target scan list sending unit, configured to send the target scan list to the server in response to the first request message; the server is used for taking the target scanning list as input data of a target wireless network positioning model to obtain a target position label of the target position to be positioned; the target wireless network positioning model is obtained for the server by adopting the following modes: obtaining a plurality of scanned list samples provided by a distribution terminal and containing wireless network samples; obtaining a scanning list feature vector for vector representation of feature information of the scanning list sample according to the hypergraph constructed by the wireless network sample and the scanning list sample; obtaining each position label sample which corresponds to the scanning time of each scanning list sample and is used for representing that the delivery capacity is positioned at a goods taking position and is provided by a delivery terminal, wherein the position label sample comprises first time period information of the delivery capacity record positioned at the goods taking position; the scanning time of the scanning list sample is in a first time period; obtaining the probability that the scanning time of the scanning list sample is in a second time period; the second time period is an actual time period in which the delivery capacity determined based on the first time period is located at the pick-up location; training an initial wireless network positioning model based on the characteristic vector of the scanning list, the probability and the position label sample to obtain a target wireless network positioning model, wherein the target wireless network positioning model is used for obtaining a position label of a position to be positioned according to a scanning list which is provided by a distribution terminal and contains a wireless network and scanned at the position to be positioned.

The present application provides an electronic device including: a processor; and the memory is used for storing a computer program which is run by the processor and executes the positioning method and the data processing method based on the wireless network data.

The present application provides a computer storage medium storing a computer program to be executed by a processor to perform the above-described positioning method and data processing method based on wireless network data.

Compared with the prior art, the embodiment of the application has the following advantages:

the application provides a positioning method based on wireless network data, because in the method, firstly, a plurality of scanned list samples which are provided by a distribution terminal and contain wireless network samples are obtained; meanwhile, obtaining each position label sample which corresponds to the scanning time of each scanning list sample and is used for indicating that the delivery capacity is positioned at the delivery position and is used for acquiring the delivery capacity record provided by the delivery terminal; then, according to the hypergraph constructed by the wireless network sample and the scanning list sample, obtaining a scanning list feature vector for vector representation of the feature information of the scanning list sample; then, obtaining the probability that the scanning time of the scanning list sample is in the second time period; actually, the probability that the scanning time of the scanning list sample is in the second time period is equivalent to correcting the position label sample reported by the delivery rider, so that the scanning list sample can be corresponding to the position label sample to which the scanning list sample truly belongs, and the target wireless network positioning model obtained by training the initial wireless network positioning model based on the scanning list feature vector, the probability and the position label sample can obtain more accurate position information, namely: the target wireless network positioning model obtained through training by the method can accurately obtain the position label of the position to be positioned according to the scanning list comprising the wireless network, which is provided by the distribution terminal and scanned at the position to be positioned. Meanwhile, based on the target wireless network positioning model, a position label of the position to be positioned can be obtained based on a scanning list containing the wireless network scanned at the position to be positioned at any moment, so that the target wireless network positioning model obtained by the method is wider in applicability.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the present application, and other drawings may be obtained according to these drawings for a person having ordinary skill in the art.

Fig. 1 is a flowchart of a positioning method based on wireless network data according to a first embodiment of the present application.

Fig. 2 is a block diagram of a training initial wireless network positioning model according to a first embodiment of the present application.

FIG. 3 is a schematic diagram of a dual-attentiveness-mechanism pre-training model based on a transducer model.

Fig. 4 is a flowchart of a data processing method according to a second embodiment of the present application.

Fig. 5 is a schematic diagram of a positioning device based on wireless network data according to a third embodiment of the present application.

Fig. 6 is a schematic diagram of a data processing apparatus according to a fourth embodiment of the present application.

Fig. 7 is a schematic diagram of an electronic device according to a fifth embodiment of the present application.

Detailed Description

In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. The present application may be embodied in many other forms than those herein described, and those skilled in the art will readily appreciate that the present application may be similarly embodied without departing from the spirit or essential characteristics thereof, and therefore the present application is not limited to the specific embodiments disclosed below.

The application provides a positioning method based on wireless network data, a data processing method, a positioning device based on wireless network data, a data processing device, an electronic device and a computer storage medium. The following describes a positioning method, a data processing method, a positioning device, a data processing device, an electronic device and a computer storage medium based on wireless network data, respectively, by specific embodiments. In order to more clearly understand the positioning method based on wireless network data provided by the embodiment of the present application, an application scenario of the positioning method based on wireless network data provided by the embodiment of the present application is first introduced.

The positioning method based on the wireless network data can be applied to a scene of indoor positioning of the delivery riders in the delivery process. In particular, the delivery rider may be in an indoor environment for a period of time during the delivery process; for example, when a business providing a meal is located in a large mall while a take-out meal is being dispensed, a dispensing rider may be in an indoor environment during the meal taking process. If the delivery rider needs to be positioned while the delivery rider is in an indoor environment, the delivery rider cannot be positioned accurately by the GPS method.

In the present application, the delivery rider in the indoor environment can be positioned based on the wireless network data. Specifically, when the delivery rider is in an indoor environment, if the delivery terminal of the delivery rider is in a function of opening the wireless network scanning, the delivery terminal can scan different available wireless networks when the delivery rider is in different indoor positions. Assume that merchant 1, merchant 2, merchant 3, and merchant 4 are present in the first layer of the mall; the wireless networks corresponding to the four merchants are a wireless network 1, a wireless network 2, a wireless network 3 and a wireless network 4, and when a distribution rider is positioned on the first layer of the mall, the distribution terminal can scan the wireless network 1, the wireless network 2, the wireless network 3 and the wireless network 4 at the same time; assume that merchants 5, 6 and 7 exist in the second layer of the mall; the wireless networks corresponding to the three merchants are a wireless network 5, a wireless network 6 and a wireless network 7, and when the distribution rider is positioned at the second layer of the mall, the distribution terminal can scan the wireless network 5, the wireless network 6 and the wireless network 7 at the same time; assume that merchants 8 and 9 exist in the second layer of the mall; the wireless networks corresponding to the two merchants are wireless networks 8 and 9, and when the distribution rider is located at the third floor of the mall, the distribution terminal can scan the wireless networks 8 and 9 at the same time.

Indoor positioning of the delivery rider can be performed based on the wireless network scanned by the delivery terminal, as the method of the present application is implemented based on the location tags (including data to the store, to take the meal, and to leave the store) reported by the delivery rider during the meal taking process and the wireless network scanned by the delivery terminal. For example, assume that a distribution rider needs to take a meal at merchant 9 and merchant 1, assume that the distribution rider reports a postprandial departure at merchant 9 at eleven (the distribution rider reports a departure at eleven for order a of merchant 9 at the distribution terminal), and quite reports a postprandial departure at merchant 1 at eleven (the distribution rider reports a departure at eleven for order B of merchant 1 at the distribution terminal). At eleven points, the scanning list which is scanned by the distribution terminal and contains the wireless network 8 and the wireless network 9; when eleven points are ten, the scanning list which is scanned by the distribution terminal and contains the wireless network comprises a wireless network 1, a wireless network 2, a wireless network 3 and a wireless network 4; if the scan list including the wireless network scanned by the one-tenth two-half distribution terminal includes the wireless network 5, the wireless network 6, and the wireless network 7, it may be determined that the distribution rider is located at the second floor of the mall. In practice, it is also possible to determine which merchant the delivery rider is located in the vicinity of based on the scanned network signal strength information of the wireless network.

Assume that the scan list including wireless network 1, wireless network 2, wireless network 3, and wireless network 4 is scan list 1; the scan list including the wireless network 5, the wireless network 6, and the wireless network 7 is the scan list 2; in the process that a delivery rider takes meals from a merchant 9 at a third layer of a market and then takes meals from a merchant 1, a first scanning list scanned by a delivery terminal is assumed to be the scanning list 3, a second scanning list is assumed to be the scanning list 2, a third scanning list is assumed to be the scanning list 1, and in fact, the scanning list 3 and the scanning list 2 are two scanning lists adjacent in time; scan list 2 and scan list 1 are two scan lists that are temporally adjacent. Through the position labels of the two adjacent scanning lists corresponding to one scanning list, the position label of the other scanning list can be determined, and then the delivery riders are positioned based on the wireless network.

The above-described example is a diagram of an application scenario of the positioning method based on wireless network data according to the present application, and the application scenario of the positioning method based on wireless network data according to the embodiment of the present application is not specifically limited, but is merely one embodiment of the application scenario of the positioning method based on wireless network data according to the present application, and the purpose of the application scenario embodiment is to facilitate understanding of the positioning method based on wireless network data according to the present application, and is not limited to the positioning method based on wireless network data according to the present application. The embodiment of the application does not need to be repeated for other application scenes of the positioning method based on the wireless network data.

First embodiment

The first embodiment of the application provides a positioning method based on wireless network data. The execution body of this embodiment is a server, and for some specific examples or details of this embodiment, please refer to the above-mentioned scenario embodiment.

The positioning method based on the wireless network data, which is provided by the embodiment of the application, is applied to the server and comprises the following steps of.

Step S101: a plurality of scanned list samples including wireless network samples provided by the distribution terminal are obtained.

In this embodiment, the indoor positioning is performed on the position to be positioned where the delivery rider is located based mainly on a scan list including a wireless network scanned by the delivery terminal of the delivery rider at the position to be positioned. The specific method can be that a target wireless network positioning model is firstly obtained through training, a scanning list which is scanned at the position to be positioned and contains the wireless network is input into the target wireless network positioning model, and then the position label of the position to be positioned can be obtained.

Specifically, please refer to fig. 2, which is a block diagram of a training initial wireless network positioning model according to a first embodiment of the present application. In the training process, the hypergraph construction module in fig. 2 is adopted to construct the hypergraph. Processing the hypergraph by adopting a topological relation extraction module in fig. 2 to obtain a matrix to be coded (see an AP topological structure vector in fig. 2) containing topological relation information between wireless network samples in the hypergraph; and finally, the position label generating module trains to obtain the target wireless network positioning model based on the position labels reported by the delivery riders, the scanning list feature vectors and the report habit data of the delivery riders. The respective information or data processing procedures in the respective modules are explained in detail below.

In the take-away delivery scenario, when the position tag of the position to be located is obtained, the position tag may refer to which merchant is in the vicinity of which merchant, and which merchant is then taken as the position tag.

In the process of training to obtain the target wireless network positioning model, a plurality of scanned list samples containing wireless network samples provided by the distribution terminal are required to be obtained, and each scanned list sample can contain a plurality of wireless network samples.

Specifically, each scan list sample is a history scan list scanned by the delivery terminal, and when the delivery rider is at a different location, a different scan list is obtained. For example, when the delivery rider is at a first floor of a mall, the delivery terminal scans a different scan list containing wireless networks than when it is at a second floor. See for a detailed description of an example scenario.

In the process of training to obtain the target wireless network positioning model, some historical scan lists can be screened as scan list samples.

After obtaining the scan list sample, a position tag sample corresponding to the scan list sample may also be obtained. And training the initial wireless network positioning model based on the scanning list sample and the corresponding position label sample so as to obtain the target wireless network positioning model.

Step S102: and obtaining a scanning list feature vector for vector representation of the feature information of the scanning list sample according to the hypergraph constructed by the wireless network sample and the scanning list sample.

In this embodiment, in the process of training the initial wireless network positioning model to obtain the target wireless network positioning model, the initial wireless network positioning model may be trained based on the scan list feature vector for vector representation of the feature information of the scan list sample.

In the present embodiment, as a hypergraph constructed from a wireless network sample and a scan list sample, one way to obtain a scan list feature vector for vector-representing feature information of the scan list sample: firstly, according to the hypergraph constructed by the wireless network sample and the scanning list sample, obtaining a wireless network characteristic vector for vector representation of characteristic information of the wireless network sample; and then, carrying out attention mechanism calculation on the wireless network feature vector to obtain a scan list feature vector for carrying out vector representation on the feature information of the scan list sample.

Specifically, according to the hypergraph constructed by the wireless network sample and the scan list sample, obtaining the wireless network feature vector for vector representation of the feature information of the wireless network sample may refer to: firstly, constructing a hypergraph representing the association relation between a wireless network sample and a scanning list sample according to the wireless network sample and the scanning list sample, wherein the vertex in the hypergraph represents the wireless network sample, and the superedge in the hypergraph represents the scanning list sample; then, according to the hypergraph, obtaining a matrix to be encoded containing topological relation information used for representing wireless network samples in the hypergraph; after the matrix to be encoded is obtained, a wireless network feature vector used for carrying out vector representation on the feature information of the wireless network sample is obtained according to the matrix to be encoded.

When the wireless network feature vector is obtained, a hypergraph can be constructed based on the wireless network sample and the scan list sample to provide a suitable data structure. The hypergraph effectively captures the characteristics of the wireless network sample by integrating the attributes and the characteristics, so that the scanning list characteristic vector can be acquired in the subsequent learning process.

In this embodiment, in order to obtain the matrix to be encoded, the method further includes: obtaining an adjacency matrix for representing the subordinate relation between the wireless network sample and the scanning list sample in the hypergraph according to the network signal intensity information of the hypergraph and the wireless network sample; in this implementation, as one way to obtain a matrix to be encoded containing information representing the topological relation between wireless network samples in a hypergraph, from the hypergraph: based on the adjacency matrix, the first degree matrix of the wireless network samples in the hypergraph, and the second degree matrix of the scan list samples in the hypergraph, a matrix to be encoded is obtained that contains information representing the topological relation between the wireless network samples in the hypergraph.

Hypergraphs are mathematical models made up of a set of vertices and a set of hyperedges that can be used to represent complex relationships between entities. Unlike conventional graphs, one hyperedge of a hypergraph may connect more than two vertices. In this embodiment, each wireless network sample may be regarded as a vertex, and each scan list sample may be regarded as a superside of the wireless network sample having an association relationship in the connection scan list samples.

Specifically, the hypergraph may be defined as a graph g= (V, S), where v= { V ₁,…,v_n } represents a vertex set (i.e., a set of wireless network samples), and s= { S ₁,…,s_m } represents a set of hyperedges (i.e., a set of scan list samples). Each superside can be connected with a plurality of vertexes, and v epsilon s and v can be usedS represents that each wireless network sample is in each scan list sample or not in each scan list sample, respectively. The structure of hypergraph G can be represented by the following adjacency matrix a _ij.

Wherein RSSIv _i,s_j represents network signal strength information between the ith wireless network sample and the jth scan list sample; i is less than or equal to n, j is less than or equal to m; RSSI, namely: RECEIVED SIGNAL STRENGTH Indication, namely: received signal strength.

The adjacency matrix is acquired so as to facilitate the subsequent extraction of the association relationship between the characteristics of the wireless network samples, wherein the association relationship between the characteristics of the wireless network samples is represented in the hypergraph through the topological relationship.

After obtaining the adjacency matrix, a matrix to be encoded containing information representing the topological relation between the wireless network samples in the hypergraph is obtained based on the adjacency matrix, the first degree matrix of the wireless network samples in the hypergraph, and the second degree matrix of the scan list samples in the hypergraph, as follows.

Firstly, defining the (SSID, BSSID) in the acquired wireless network sample data as the ID (Identity Document, i.e. identification) of the wireless network sample to distinguish different wireless network samples, encoding the characteristic information of each wireless network sample into a D-dimensional potential space tensor (the characteristic information of each wireless network sample includes the association relationship between the characteristics of the wireless network sample), and forming a matrix to be encoded (the matrix to be encoded, i.e. the initial embedding representation of the characteristics of the wireless network sample, and each row in the matrix to be encoded may represent the initial embedding of the characteristics of one wireless network sample) by the corresponding D-dimensional potential space tensor of all the wireless network samples. The data for each wireless network sample contains SSID (SERVICE SET IDENTIFIER: service set identification), BSSID (Basic SERVICE SET IDENTIFIER: basic service set identification) and RSSI. The SSID technology can divide a wireless local area network into a plurality of sub-networks requiring different identity verification, each sub-network needs independent identity verification, and only users passing the identity verification can enter the corresponding sub-network to prevent unauthorized users from entering the network. The BSSID refers to a basic service set identifier in the wireless local area network, which is a unique identifier for identifying each wireless access point in the wireless local area network, similar to a MAC address.

Considering that the topology between wireless network samples (each wireless network sample may be represented by one AP, i.e., v _n is a wireless network sample generated by AP _n, AP Access Poin is a wireless access point, is an access point of a wireless network, is commonly referred to as a "hot spot", i.e., indicates that the wireless network sample is a wireless network generated by the AP) may be helpful for indoor positioning, and thus hypergraph information is used to capture the topology between APs (the topology between APs indicates the topology between wireless network samples).

By encoding adjacent sub-graph structures in the hypergraph, contextual information and interdependencies in the hypergraph can be more effectively captured, which will enable a better understanding of the topology of the hypergraph later and increase the ability to handle relationships between APs. The matrix to be encoded can be calculated by a laplacian matrix, and the hypergraph laplacian matrix L after symmetric standardization is as follows.

Where I is the identity matrix, D _v and D _s represent the degree matrices of nodes and edges, respectively, in the hypergraph, and A represents the adjacency matrix of the hypergraph. Because the adjacency matrix of the hypergraph contains the RSSI value, the matrix to be coded calculated based on the adjacency matrix not only can learn the connection relation between the APs (the connection relation is embodied by the topological relation), but also can take the influence of different distances between the APs (the distance mainly refers to the distance between the APs and the distribution terminal, and the distance is positively related to the RSSI value) into consideration.

In a word, the topological relation between APs (the topological relation between APs represents the association relation between the features of the wireless network samples) is obtained by analyzing the hypergraph, and a more meaningful feature space of the wireless network samples can be constructed through the topological relation between the APs, so that subsequent learning is enhanced, and the feature vector of the scanning list is obtained conveniently.

The subsequent learning process corresponds in effect to a process of performing an attention mechanism calculation on the wireless network feature vector to obtain a scan list feature vector for vector representation of feature information of the scan list sample. The learning process is responsible for learning global perceptual features related to a specific pattern in the wireless network samples (the global perceptual features related to a specific pattern may refer to association relationships among a plurality of wireless network samples in a certain scan list sample), and the global perceptual features are not directly output by the encoder but are in parameters of the encoder itself; potential features are extracted from the wireless network sample and the scan list sample through a dual-attention mechanism, and effective characterization of the scan list sample is obtained.

For merchant-level indoor positioning, one key objective is to obtain a meaningful representation of a scan list sample containing wireless network samples. Learning a particular pattern between the wireless network sample and the scan list sample will help obtain a meaningful representation of the scan list sample (a meaningful representation of the scan list sample, i.e., a scan list feature vector).

Specifically, performing attention mechanism calculation on the wireless network feature vector to obtain a scan list feature vector for vector representation of feature information of a scan list sample may refer to: firstly, encoding a wireless network feature vector by adopting a first attention mechanism to obtain an encoded wireless network feature vector; and then, aggregating the encoded wireless network feature vectors by adopting a second attention mechanism to obtain a scanning list feature vector for vector representation of the feature information of the scanning list sample.

The existing BERT network model (Bidirectional Encoder Representations from Transformers, namely a bi-directional coding pre-training natural language processing model) has good effect in learning space-time characteristics and position related information; in the application, a model which is more suitable for wireless network data is explored on the basis, and because the sequence of wireless network samples in the scanned list samples obtained by scanning is random and information related to the position is embedded in RSSI values, potential information which is irrelevant to the sequence and influences each other among the wireless network samples in the scanned list samples and the relation between the RSSI values and the positioning of the wireless network samples are particularly focused in the learning process of the application, a dual-attention mechanism pre-training model based on a converter model (namely a converter) is provided in the application, and is a schematic diagram of the dual-attention mechanism pre-training model based on the converter model as shown in figure 3.

The converter model based dual-attentiveness mechanism pre-training model includes SCANLIST SELF-attention (scan list self-attentiveness mechanism, i.e., a first attentiveness mechanism for learning order-independent interaction potential information between APs within a scan list sample) and RSSI Edge attention (side-attentiveness mechanism for received signal strength, i.e., a second attentiveness mechanism, also referred to as RSSI side-attentiveness mechanism, for capturing potential characteristics between RSSI values and location).

Specifically, the wireless network feature vector is encoded by adopting a first attention mechanism, and the obtained encoded wireless network feature vector may be: and encoding the association relation characteristic among the wireless network samples in the scanning list samples into the wireless network characteristic vector by adopting a preset encoder to obtain the encoded wireless network characteristic vector.

As shown in fig. 3, the AP-corresponding features are updated by the scan list self-attention mechanism, and in fact, when the first attention mechanism is used for encoding, new information is continuously encoded based on the initial embedding of the features of the wireless network samples, so that new information is continuously encoded, because some feature information is lost in the initial embedding of the features of each wireless network sample when the matrix to be encoded is generated. The predetermined encoder is such as the transducer model encoder of fig. 3 (i.e., transformer Encoder).

Since the order in which the wireless network samples appear in the scan list samples is generally arbitrary, if it is desired that the results output by the converter model encoder remain unchanged at any permutation of the wireless network samples, no positional embedding can be employed as input to the converter model encoder to infer order-independent dependencies between APs in different semantic contexts. Given an AP topology vector (e.g., AP ₁ -AP _n in fig. 3), the converter model encoder updates the matrix (i.e., encodes the matrix to be encoded) by a multi-headed self-care mechanism. At the same time, FFN (Feed-forward Network) and LN (Layer Normalization ) methods with a ReLU (RECTIFIED LINEAR Unit) activated linear transform layer can also be utilized to improve the ability of the converter model encoder to extract features.

After the encoded wireless network feature vectors are obtained (e.g., e ₁ to e _n,e₁ to e _n in fig. 3 are the encoded wireless network feature vectors used to represent the first wireless network sample to the nth wireless network sample, respectively), the encoded wireless network feature vectors are aggregated using a second attention mechanism to obtain a scan list feature vector used to vector-represent the feature information of the scan list sample.

Specifically, aggregating the encoded wireless network feature vectors by adopting a second attention mechanism to obtain a scan list feature vector for vector representation of feature information of a scan list sample, which may refer to: firstly, taking network signal intensity information of a wireless network sample as a weight to weight the encoded wireless network feature vector, and obtaining a weighted wireless network feature vector; and then, aggregating the weighted wireless network feature vectors to obtain a scan list feature vector for vector representation of the feature information of the scan list sample.

The above procedure is actually to aggregate the feature information of a plurality of wireless network samples contained in the scan list sample together by a side-note force mechanism of the received signal strength, so as to obtain a scan list feature vector for vector representation of the feature information of the scan list sample.

In particular, the RSSI values contain a lot of location-related information due to indoor positioning problems for wireless networks. In the delivery scenario, the closer the delivery rider is to the AP, the higher the corresponding RSSI value, and the greater the impact on the delivery rider's positioning location. Therefore, the wireless network characteristic vector coded by each wireless network sample can be spliced together after being multiplied by the corresponding RSSI. It should be noted that, RSSI may be used as a trainable weight matrix to adapt to different scenarios. The vector dimension can be changed by multiplying each wireless network sample encoded wireless network feature vector with the corresponding RSSI and then stitching together. For example, when the dimension of the weight matrix corresponding to the RSSI is E-dimension, the wireless network feature vector (e.g., D-dimension) after each wireless network sample is encoded is multiplied by the corresponding RSSI and then spliced together, so as to obtain the scan list feature vector in E-dimension.

Specifically, aggregating the weighted wireless network feature vectors to obtain a scan list feature vector for vector representation of feature information of the scan list sample may be as follows.

First, a scan list sample of any two adjacent scan times in a plurality of scan list samples is determined.

And then, based on the scan list samples of two adjacent scan times, aggregating the weighted wireless network feature vectors to obtain a scan list feature vector for vector representation of the feature information of the scan list samples.

In fact, in aggregating with the second attention mechanism, the process also needs to be trained so that the second attention mechanism can determine scan list samples for any two adjacent scan times in the plurality of scan list samples. For example, referring to fig. 3, e ₁ to e _i may be aggregated as S ₁; the scan list feature vectors of the scan list samples of two adjacent scan times can be aggregated into S ₂,S₁ and S ₂ based on e _i+1 to e _n. Of course, scan list feature vectors for m scan list samples may ultimately be obtained based on the second attention mechanism.

Step S103: and obtaining each position label sample which corresponds to the scanning time of each scanning list sample and is used for indicating that the delivery capacity is positioned at the delivery position of the delivery terminal.

In this embodiment, the initial wireless network positioning model is trained based on the scan list sample and the position label sample corresponding to the scan list sample, so that each position label sample for indicating that the delivery capacity is located at the pick-up position, which corresponds to the scan time of each scan list sample, of the delivery capacity record provided by the delivery terminal can be obtained.

In a delivery scenario, a sample of location tags is such as historical location tags reported by delivery riders (including historical time data for arrival, fetch, and departure reports). Of course, the location tag sample contains location information.

In this embodiment, the location tag sample includes first time period information of the delivery capacity record at the pick-up location; the scan time of the scan list sample is within a first time period.

For example, delivery rider a records eight to eight points quite within merchant M (which may be the assignment of delivery rider a reporting eight points to the location of merchant M, eight points quite reporting the location of completed meal removal from merchant M); the scan time of the scan list sample P is one-minute at eight points, and the scan time of the scan list sample P is in a period of between eight points and ten minutes at eight points.

Step S104: a probability is obtained that the scan time of the scan list sample is within the second time period.

In this embodiment, the second time period is an actual time period in which the delivery capacity determined based on the first time period is located at the pick-up location.

In the actual delivery process, the delivery rider may have false alarms or data that is used to report to the store, get meals and leave the store in advance, so that the obtained position tag sample recorded by the delivery rider does not accord with the recorded time period. For example, assuming that delivery rider a is accustomed to reporting to the store two minutes in advance (the reporting habit data of delivery rider a is reported two minutes in advance), the corresponding delivery rider a is actually two to eight minutes at the time period of the merchant M. Namely: in this example, the second period is eight-point two to eight-point ten, and in practice, the second period is a correction period for the first period. In this case, it is equivalent to the need to correct the position tag sample recorded by the delivery rider for the scan list sample, and the correction process is related to the probability obtained as described above.

In fact, in step S103, if each of the position tag samples corresponding to each of the scan list samples of the delivery capacity record is accurate (for example, the delivery rider can accurately report the position), no correction is required.

To correct the position tag samples for the scan list samples, a probability that the scan time of the scan list samples is within a second time period may be obtained, and the correction based on the obtained probability.

In this embodiment, the first time period information includes first time information of arrival at the pick-up location recorded by the delivery capacity, and second time information of departure from the pick-up location recorded by the delivery capacity.

In this embodiment, further comprising: the second time period corresponding to the first time period is determined in the following specific way: according to the first time information and the second time information, third time information that the delivery capacity actually reaches the pick-up position and fourth time information that the delivery capacity actually leaves the pick-up position are obtained; and determining a second time period according to the third time information and the fourth time information.

In this embodiment, further comprising: acquiring first time error information of delivery capacity record reaching a goods taking position and second time error information of delivery capacity record leaving the goods taking position; the obtaining of the third time information that the delivery capacity actually reaches the pick-up location and the fourth time information that the delivery capacity actually leaves the pick-up location according to the first time information and the second time information may refer to: obtaining third time information according to the first time information and the first time error information; and obtaining fourth time information according to the second time information and the second time error information.

For example, when the report habit data of the delivery rider a is reported to the store in advance and the time error information is reported to the store in advance by two minutes, the third time information (eight minutes) may be obtained based on the first time information (eight minutes) and the first time error information (two minutes in advance); the fourth time information (eight-point ten) may be obtained based on the second time information (eight-point ten) and the second time error information (assuming that the delivery rider a is accustomed to reporting departure data).

As one way of obtaining the probability that the scan time of the scan list sample is within the second time period: taking the scanning list sample as input data of a scanning list probability acquisition model to acquire the probability that the scanning time of the scanning list sample is in a second time period; the scan list probability obtaining model is used for obtaining the probability that the scan time of the scan list is in the actual time period that the delivery capacity is located at the pick-up position corresponding to the scan list according to the scan list.

The scan list samples within the real time [ arrival time, departure time ] interval of the delivery rider to the store may be marked as positive samples for the corresponding merchant. However, since there are a lot of errors in the position tags manually reported by the delivery rider, they cannot be used directly, and thus fine-grained statistics are used to estimate the period of time that the delivery rider actually reaches the merchant. The specific method comprises the following steps: the probability of the scan time of a certain scan list sample in the time interval between the moment T _a 'when the delivery rider actually arrives at the merchant to the moment T _d' when the delivery rider actually leaves the merchant is calculated based on the moment T _a when the delivery rider reports to the merchant, the probability distribution of the actual arrival moment and the report arrival moment being f _a, the moment T _d when the delivery rider reports to leave the merchant, and the probability distribution of the actual departure moment and the report departure moment being f _d, which can be expressed specifically as the following formula.

Where T is the current time between T _a ' and T _d ', and DeltaT _a is the time difference between T _a ' and T _a; ΔT _d is the time difference between T _d' and T _d; max represents the maximum time difference (Max is determined according to the strategy in the delivery domain) that the delivery rider can report to the pick-up location in advance or in retard; min represents the minimum time difference (which may be 0) that the delivery rider may report off the pickup location in advance or in retard. The value range of k in the first sigma in the formula is T-T _a to Max; the value of k in the second sigma ranges from Min to T-T _d.

The probability formula corresponds to an example of the scan list probability acquisition model.

Step S105: training the initial wireless network positioning model based on the scanning list feature vector, the probability and the position label sample to obtain a target wireless network positioning model.

After calculating the probabilities, the reported position tag samples may be corrected based on the probabilities and the corresponding reported position tag samples. For example, when the scan time of the scan list sample P is one-eighth, assuming that the delivery rider a records that the time of leaving the merchant N is seventy-ninth, since the delivery rider actually reports that the time of leaving the merchant N is two minutes earlier in order to report that the arrival at the merchant M, it can be confirmed that the true position tag sample of the scan list sample P may be the merchant N by calculating the above probability. In practice, when training the initial wireless network positioning model based on the calculated probability and the position label sample of the corresponding report, the position label sample after correction is used as a training sample for training. The corrected position tag samples are in fact theoretical position tag samples.

And training the initial wireless network positioning model based on the scanning list feature vector and the corresponding corrected position label sample to obtain a target wireless network positioning model.

The initial wireless network positioning model may actually include a dual-attentiveness mechanism pre-training model based on a transducer model, a scan list probability obtaining model and an initial position tag generating model, and in step S105, the initial position tag generating model is mainly trained.

In this embodiment, the target wireless network positioning model is configured to obtain a position tag of a to-be-positioned position according to a scan list including wireless networks scanned at the to-be-positioned position provided by the distribution terminal.

In this embodiment, further comprising: obtaining a target scanning list which is provided by a distribution terminal and scanned at a target to-be-positioned position and contains a target wireless network; and taking the target scanning list as input data of a target wireless network positioning model to obtain a target position label of a target position to be positioned.

Second embodiment

Corresponding to the first embodiment, a second embodiment of the present application provides a data processing method. The execution body of this embodiment is a distribution terminal, and the same parts as those of the first embodiment in the second embodiment will not be described, specifically please refer to the relevant parts of the first embodiment and the scene embodiment.

The data processing method of the embodiment of the application is applied to the distribution terminal and comprises the following steps.

Step S401: and obtaining a first request message which is sent by the server and used for requesting to obtain a target scanning list which is scanned at the target to-be-positioned position and contains the target wireless network.

Step S402: and responding to the first request message, and sending a target scanning list to the server side.

In this embodiment, the server is configured to use the target scan list as input data of a target wireless network positioning model to obtain a target position tag of a target position to be positioned; the target wireless network positioning model is obtained for the server side in the following way: obtaining a plurality of scanned list samples provided by a distribution terminal and containing wireless network samples; obtaining a scanning list feature vector for vector representation of feature information of the scanning list sample according to the hypergraph constructed by the wireless network sample and the scanning list sample; obtaining each position label sample which corresponds to the scanning time of each scanning list sample and is used for indicating that the delivery capacity is positioned at the delivery position and is provided by the delivery terminal, wherein the position label sample comprises first time period information of the delivery capacity record positioned at the delivery position; the scanning time of the scanning list sample is within a first time period; obtaining the probability that the scanning time of the scanning list sample is in the second time period; the second time period is an actual time period in which the delivery capacity determined based on the first time period is located at the pickup position; training an initial wireless network positioning model based on the characteristic vector of the scanning list, the probability and the position label sample to obtain a target wireless network positioning model, wherein the target wireless network positioning model is used for obtaining a position label of a position to be positioned according to a scanning list which is provided by a distribution terminal and contains a wireless network and scanned at the position to be positioned.

Third embodiment

Corresponding to the positioning method based on wireless network data provided in the first embodiment of the present application, the third embodiment of the present application further provides a positioning device based on wireless network data. Since the device embodiment is substantially similar to the first embodiment, the description is relatively simple, and reference is made to the partial description of the first embodiment for relevant points. The device embodiments described below are merely illustrative.

The positioning device 500 based on wireless network data is applied to a server, and the device comprises: a scan list sample obtaining unit 501, configured to obtain a plurality of scan list samples including a wireless network sample, which are provided by a distribution terminal and are scanned; wherein each scan list sample comprises a plurality of wireless network samples; a scan list feature vector obtaining unit 502, configured to obtain a scan list feature vector for vector representation of feature information of the scan list sample according to a hypergraph constructed by the wireless network sample and the scan list sample; a position tag sample obtaining unit 503, configured to obtain each position tag sample, corresponding to a scanning time of each scanning list sample, of a delivery capacity record provided by a delivery terminal, where the position tag sample is used to indicate that a delivery capacity is located at a pickup position, and the position tag sample includes first time period information of the delivery capacity record located at the pickup position; the scanning time of the scanning list sample is in a first time period; a probability obtaining unit 504, configured to obtain a probability that a scanning time of the scan list sample is within a second period of time; the second time period is an actual time period in which the delivery capacity determined based on the first time period is located at the pick-up location; the target wireless network positioning model obtaining unit 505 is configured to train an initial wireless network positioning model based on the feature vector of the scan list, the probability and the position label sample, to obtain a target wireless network positioning model, where the target wireless network positioning model is configured to obtain a position label of a to-be-positioned position according to a scan list including wireless networks scanned at the to-be-positioned position provided by a distribution terminal.

Optionally, the scan list feature vector obtaining unit is specifically configured to: obtaining a wireless network feature vector for vector representation of feature information of the wireless network sample according to the hypergraph constructed by the wireless network sample and the scanning list sample; and performing attention mechanism calculation on the wireless network feature vector to obtain a scan list feature vector for vector representation of the feature information of the scan list sample.

Optionally, the scan list feature vector obtaining unit is specifically configured to: constructing a hypergraph representing the association relationship between the wireless network sample and the scanning list sample according to the wireless network sample and the scanning list sample, wherein vertexes in the hypergraph represent the wireless network sample, and superedges in the hypergraph represent the scanning list sample; obtaining a matrix to be encoded containing topological relation information used for representing the wireless network samples in the hypergraph according to the hypergraph; and obtaining a wireless network characteristic vector for carrying out vector representation on the characteristic information of the wireless network sample according to the matrix to be encoded.

Optionally, the method further comprises: an adjacency matrix obtaining unit configured to obtain an adjacency matrix for representing a relationship between the wireless network samples and the scan list samples in the hypergraph, based on the network signal strength information of the hypergraph and the wireless network samples; the scan list feature vector obtaining unit is specifically configured to: based on the adjacency matrix, a first degree matrix of the wireless network samples in the hypergraph, and a second degree matrix of the scan list samples in the hypergraph, a matrix to be encoded is obtained that includes information representing a topological relationship between the wireless network samples in the hypergraph.

Optionally, the scan list feature vector obtaining unit is specifically configured to: encoding the wireless network feature vector by adopting a first attention mechanism to obtain an encoded wireless network feature vector; and aggregating the encoded wireless network feature vectors by adopting a second attention mechanism to obtain a scanning list feature vector for vector representation of the feature information of the scanning list sample.

Optionally, the scan list feature vector obtaining unit is specifically configured to: and encoding the association relation characteristic among the wireless network samples in the scanning list samples into the wireless network characteristic vector by adopting a preset encoder to obtain an encoded wireless network characteristic vector.

Optionally, the scan list feature vector obtaining unit is specifically configured to: weighting the coded wireless network feature vector by taking the network signal intensity information of the wireless network sample as a weight to obtain a weighted wireless network feature vector; and aggregating the weighted wireless network feature vectors to obtain a scanning list feature vector for vector representation of the feature information of the scanning list sample.

Optionally, the scan list feature vector obtaining unit is specifically configured to: determining a scan list sample of any two adjacent scan times of the plurality of scan list samples; and aggregating the weighted wireless network feature vectors based on the scan list samples of the two adjacent scan times to obtain a scan list feature vector for vector representation of the feature information of the scan list samples.

Optionally, the probability obtaining unit is specifically configured to: taking the scanning list sample as input data of a scanning list probability acquisition model to acquire the probability that the scanning time of the scanning list sample is in a second time period; the scan list probability obtaining model is used for obtaining the probability that the scan time of the scan list is in the actual time period that the delivery capacity is located at the pick-up position corresponding to the scan list according to the scan list.

Optionally, the first time period information includes first time information of arrival at the pick-up location recorded by delivery capacity, second time information of departure from the pick-up location recorded by delivery capacity; the apparatus further comprises: the second time period determining unit is specifically configured to: according to the first time information and the second time information, third time information that the delivery capacity actually reaches the goods taking position and fourth time information that the delivery capacity actually leaves the goods taking position are obtained; and determining the second time period according to the third time information and the fourth time information.

Optionally, the method further comprises: the error information obtaining unit is used for obtaining first time error information of the delivery capacity record reaching the goods taking position and second time error information of the delivery capacity record leaving the goods taking position; the second time period determining unit is specifically configured to: obtaining the third time information according to the first time information and the first time error information; and obtaining the fourth time information according to the second time information and the second time error information.

Optionally, the method further comprises: the target position label obtaining unit is specifically configured to: obtaining a target scanning list which is provided by the distribution terminal and scanned at a target to-be-positioned position and contains a target wireless network; and taking the target scanning list as input data of the target wireless network positioning model to obtain a target position label of the target position to be positioned.

Fourth embodiment

The fourth embodiment of the present application also provides a data processing apparatus corresponding to the data processing method provided by the second embodiment of the present application. Since the device embodiment is substantially similar to the second embodiment, the description is relatively simple, and reference is made to the description of the second embodiment for relevant points. The device embodiments described below are merely illustrative.

Fig. 6 is a schematic diagram of a data processing apparatus according to a fourth embodiment of the application.

The data processing apparatus 600 is applied to a distribution terminal, and the apparatus includes: a first request message obtaining unit 601, configured to obtain a first request message sent by a server and used for requesting to obtain a target scan list including a target wireless network scanned at a target to-be-located position; a target scan list sending unit 602, configured to send the target scan list to the server in response to the first request message; the server is used for taking the target scanning list as input data of a target wireless network positioning model to obtain a target position label of the target position to be positioned; the target wireless network positioning model is obtained for the server by adopting the following modes: obtaining a plurality of scanned list samples provided by a distribution terminal and containing wireless network samples; obtaining a scanning list feature vector for vector representation of feature information of the scanning list sample according to the hypergraph constructed by the wireless network sample and the scanning list sample; obtaining each position label sample which corresponds to the scanning time of each scanning list sample and is used for representing that the delivery capacity is positioned at a goods taking position and is provided by a delivery terminal, wherein the position label sample comprises first time period information of the delivery capacity record positioned at the goods taking position; the scanning time of the scanning list sample is in a first time period; obtaining the probability that the scanning time of the scanning list sample is in a second time period; the second time period is an actual time period in which the delivery capacity determined based on the first time period is located at the pick-up location; training an initial wireless network positioning model based on the characteristic vector of the scanning list, the probability and the position label sample to obtain a target wireless network positioning model, wherein the target wireless network positioning model is used for obtaining a position label of a position to be positioned according to a scanning list which is provided by a distribution terminal and contains a wireless network and scanned at the position to be positioned.

Fifth embodiment

The fifth embodiment of the present application also provides an electronic device corresponding to the methods of the first and second embodiments of the present application.

Fig. 7 is a schematic diagram of an electronic device according to a fifth embodiment of the present application, as shown in fig. 7.

In this embodiment, an optional hardware structure of the electronic device 700 may be as shown in fig. 7, including: at least one processor 701, at least one memory 702, and at least one communication bus 705; the memory 702 contains a program 703 and data 704.

Bus 705 may be a communication device that transfers data between components within electronic device 700, such as an internal bus (e.g., a CPU-memory bus, processor central processing unit, CPU for short), an external bus (e.g., a universal serial bus port, a peripheral component interconnect express port), and so forth.

In addition, the electronic device further includes: at least one network interface 706, at least one peripheral interface 707. The network interface 706 to provide wired or wireless communication with an external network 708 (e.g., the Internet, an intranet, a local area network, a mobile communication network, etc.); in some embodiments, the network interface 706 may include any number of network interface controllers (English: network interface controller, NIC for short), radio Frequency (RF) modules, transponders, transceivers, modems, routers, gateways, any combination of wired network adapters, wireless network adapters, bluetooth adapters, infrared adapters, near Field Communication (NFC) adapters, cellular network chips, and the like.

Peripheral interface 707 is used to connect with peripherals, such as peripheral 1 (709 in fig. 7), peripheral 2 (710 in fig. 7), and peripheral 3 (711 in fig. 7). Peripherals, i.e., peripheral devices, which may include, but are not limited to, cursor control devices (e.g., mice, touchpads, or touchscreens), keyboards, displays (e.g., cathode ray tube displays, liquid crystal displays). A display or light emitting diode display, a video input device (e.g., a video camera or an input interface communicatively coupled to a video archive), etc.

The processor 701 may be a CPU, or an Application-specific integrated Circuit ASIC (Application SPECIFIC INTEGRATED circuits), or one or more integrated circuits configured to implement embodiments of the present application.

The memory 702 may comprise high-speed RAM (collectively referred to as Random Access Memory a random access memory) memory, and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.

The processor 701 invokes programs and data stored in the memory 702 to execute the methods according to the first and second embodiments of the present application.

Sixth embodiment

The sixth embodiment of the present application also provides a computer storage medium storing a computer program corresponding to the methods of the first and second embodiments of the present application, the computer program being executed by a processor to perform the methods of the first and second embodiments of the present application.

While the application has been described in terms of preferred embodiments, it is not intended to be limiting, but rather, it will be apparent to those skilled in the art that various changes and modifications can be made herein without departing from the spirit and scope of the application as defined by the appended claims.

In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory. The Memory may include volatile Memory, random Access Memory (RAM), and/or nonvolatile Memory in a computer-readable medium, such as Read-Only Memory (ROM) or flash RAM. Memory is an example of computer-readable media.

1. Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change Memory (English: PHASE CHANGE Memory; PRAM), static random access Memory (English: static Random Access Memory; SRAM), dynamic random access Memory (English: dynamic Random Access Memory; DRAM), other types of Random Access Memory (RAM), read-Only Memory (ROM), electrically erasable programmable read-Only Memory (EEPROM), flash Memory or other Memory technology, read-Only optical disk read-Only Memory (English: compact Disc Read-Only Memory; CD-ROM), digital versatile disk (English: DIGITAL VERSATILE DISC; DVD) or other optical storage, magnetic cassettes, magnetic tape storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include non-transitory computer-readable storage media (non-transitory computer readable storage media), such as modulated data signals and carrier waves.

2. It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region, and provide corresponding operation entries for the user to select authorization or rejection.

Claims

1. A positioning method based on wireless network data, applied to a server, the method comprising:

Obtaining a plurality of scanned list samples provided by a distribution terminal and containing wireless network samples; wherein each scan list sample comprises a plurality of wireless network samples;

Obtaining a scanning list feature vector for vector representation of feature information of the scanning list sample according to the hypergraph constructed by the wireless network sample and the scanning list sample;

obtaining each position label sample which corresponds to the scanning time of each scanning list sample and is used for representing that the delivery capacity is positioned at a goods taking position and is provided by a delivery terminal, wherein the position label sample comprises first time period information of the delivery capacity record positioned at the goods taking position; the scanning time of the scanning list sample is in a first time period;

obtaining the probability that the scanning time of the scanning list sample is in a second time period; the second time period is an actual time period in which the delivery capacity determined based on the first time period is located at the pick-up location;

Training an initial wireless network positioning model based on the characteristic vector of the scanning list, the probability and the position label sample to obtain a target wireless network positioning model, wherein the target wireless network positioning model is used for obtaining a position label of a position to be positioned according to a scanning list which is provided by a distribution terminal and contains a wireless network and scanned at the position to be positioned.

2. The method according to claim 1, wherein the obtaining a scan list feature vector for vector representation of feature information of the scan list sample from the hypergraph constructed by the wireless network sample and the scan list sample comprises:

obtaining a wireless network feature vector for vector representation of feature information of the wireless network sample according to the hypergraph constructed by the wireless network sample and the scanning list sample;

and performing attention mechanism calculation on the wireless network feature vector to obtain a scan list feature vector for vector representation of the feature information of the scan list sample.

3. The method according to claim 2, wherein the obtaining a wireless network feature vector for vector representing feature information of the wireless network sample from the hypergraph constructed by the wireless network sample and the scan list sample includes:

Constructing a hypergraph representing the association relationship between the wireless network sample and the scanning list sample according to the wireless network sample and the scanning list sample, wherein vertexes in the hypergraph represent the wireless network sample, and superedges in the hypergraph represent the scanning list sample;

Obtaining a matrix to be encoded containing topological relation information used for representing the wireless network samples in the hypergraph according to the hypergraph;

And obtaining a wireless network characteristic vector for carrying out vector representation on the characteristic information of the wireless network sample according to the matrix to be encoded.

4. A method according to claim 3, further comprising: obtaining an adjacency matrix for representing the subordinate relation between the wireless network sample and the scanning list sample in the hypergraph according to the network signal intensity information of the hypergraph and the wireless network sample;

the obtaining, according to the hypergraph, a matrix to be encoded including information representing a topological relation between the wireless network samples in the hypergraph, including:

Based on the adjacency matrix, a first degree matrix of the wireless network samples in the hypergraph, and a second degree matrix of the scan list samples in the hypergraph, a matrix to be encoded is obtained that includes information representing a topological relationship between the wireless network samples in the hypergraph.

5. The method of claim 2, wherein said performing an attention mechanism calculation on said wireless network feature vector to obtain a scan list feature vector for vector representation of feature information of said scan list sample, comprises:

Encoding the wireless network feature vector by adopting a first attention mechanism to obtain an encoded wireless network feature vector;

And aggregating the encoded wireless network feature vectors by adopting a second attention mechanism to obtain a scanning list feature vector for vector representation of the feature information of the scanning list sample.

6. The method of claim 5, wherein encoding the wireless network feature vector using the first attention mechanism to obtain an encoded wireless network feature vector comprises:

And encoding the association relation characteristic among the wireless network samples in the scanning list samples into the wireless network characteristic vector by adopting a preset encoder to obtain an encoded wireless network characteristic vector.

7. The method of claim 5, wherein aggregating the encoded wireless network feature vectors using a second attention mechanism to obtain a scan list feature vector for vector representation of feature information of the scan list samples, comprising:

Weighting the coded wireless network feature vector by taking the network signal intensity information of the wireless network sample as a weight to obtain a weighted wireless network feature vector;

And aggregating the weighted wireless network feature vectors to obtain a scanning list feature vector for vector representation of the feature information of the scanning list sample.

8. The method of claim 7, wherein aggregating the weighted wireless network feature vectors to obtain a scan list feature vector for vector representation of feature information of the scan list samples, comprises:

determining a scan list sample of any two adjacent scan times of the plurality of scan list samples;

And aggregating the weighted wireless network feature vectors based on the scan list samples of the two adjacent scan times to obtain a scan list feature vector for vector representation of the feature information of the scan list samples.

9. The method of claim 1, wherein obtaining the probability that the scan time of the scan list sample is within a second time period comprises:

taking the scanning list sample as input data of a scanning list probability acquisition model to acquire the probability that the scanning time of the scanning list sample is in a second time period; the scan list probability obtaining model is used for obtaining the probability that the scan time of the scan list is in the actual time period that the delivery capacity is located at the pick-up position corresponding to the scan list according to the scan list.

10. The method of claim 1, wherein the first time period information comprises first time information of arrival at the pick-up location recorded by delivery capacity, second time information of departure from the pick-up location recorded by delivery capacity;

The method further comprises the steps of:

according to the first time information and the second time information, third time information that the delivery capacity actually reaches the goods taking position and fourth time information that the delivery capacity actually leaves the goods taking position are obtained;

And determining the second time period according to the third time information and the fourth time information.

11. The method as recited in claim 10, further comprising: acquiring first time error information of delivery capacity record reaching a goods taking position and second time error information of delivery capacity record leaving the goods taking position;

The obtaining, according to the first time information and the second time information, third time information that the delivery capacity actually reaches the pick-up location and fourth time information that the delivery capacity actually leaves the pick-up location includes:

obtaining the third time information according to the first time information and the first time error information;

And obtaining the fourth time information according to the second time information and the second time error information.

12. The method as recited in claim 1, further comprising:

Obtaining a target scanning list which is provided by the distribution terminal and scanned at a target to-be-positioned position and contains a target wireless network;

and taking the target scanning list as input data of the target wireless network positioning model to obtain a target position label of the target position to be positioned.

13. A data processing method applied to a distribution terminal, the method comprising:

obtaining a first request message sent by a server side and used for requesting to obtain a target scanning list which is scanned at a target to-be-positioned position and contains a target wireless network;

responding to the first request message, and sending the target scanning list to the server; the server is used for taking the target scanning list as input data of a target wireless network positioning model to obtain a target position label of the target position to be positioned; the target wireless network positioning model is obtained for the server by adopting the following modes: obtaining a plurality of scanned list samples provided by a distribution terminal and containing wireless network samples; obtaining a scanning list feature vector for vector representation of feature information of the scanning list sample according to the hypergraph constructed by the wireless network sample and the scanning list sample; obtaining each position label sample which corresponds to the scanning time of each scanning list sample and is used for representing that the delivery capacity is positioned at a goods taking position and is provided by a delivery terminal, wherein the position label sample comprises first time period information of the delivery capacity record positioned at the goods taking position; the scanning time of the scanning list sample is in a first time period; obtaining the probability that the scanning time of the scanning list sample is in a second time period; the second time period is an actual time period in which the delivery capacity determined based on the first time period is located at the pick-up location; training an initial wireless network positioning model based on the characteristic vector of the scanning list, the probability and the position label sample to obtain a target wireless network positioning model, wherein the target wireless network positioning model is used for obtaining a position label of a position to be positioned according to a scanning list which is provided by a distribution terminal and contains a wireless network and scanned at the position to be positioned.