Data processing method, device, electronic device and storage medium

Abstract

The embodiments of the present application disclose a data processing method, device, electronic device and storage medium, including: collecting longitude and latitude information and wireless signal access information of a target device; calculating a first probability corresponding to each candidate area in a preset candidate area set of the target device according to the longitude and latitude information and the wireless signal access information; selecting a target area from multiple candidate areas according to the first probability, the target area including at least one target sub-area; calling wireless signal connection information of the target area, the wireless signal connection information carrying data generated by connecting to wireless resources within a historical period; calculating a second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information; obtaining the position of the target device according to the first probability and the second probability. This scheme can improve the accuracy of positioning.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a data processing method, device, electronic device and storage medium. Background Art

[0002] At present, in indoor environments, special indoor positioning technology is needed to achieve indoor positioning. Indoor positioning refers to the realization of position positioning in indoor environments. It mainly uses wireless communication, base station positioning, inertial navigation positioning and other technologies to integrate a set of indoor positioning systems, thereby realizing the position detection of people, objects, etc. in indoor spaces.

[0003] At present, there are two indoor positioning solutions: one uses radio frequency technology for indoor positioning, and the other uses longitude and latitude information for indoor positioning. However, the first solution has high requirements on equipment and cannot match wide-area candidate areas. When the candidate area is very large, precise positioning requires multiple repeated calculations, which has a high computational overhead. In the second solution, the signal is easily affected by wall obstructions, resulting in large errors in longitude and latitude, making it impossible to perform precise positioning. Summary of the invention

[0004] The embodiments of the present invention provide a data processing method, device, electronic device and storage medium, which can reduce the amount of calculation during positioning and improve the accuracy of positioning.

[0005] An embodiment of the present invention provides a data processing method, including:

[0006] Collect the latitude and longitude information of the target device and the wireless signal access information;

[0007] Calculating, according to the latitude and longitude information and the wireless signal access information, a first probability corresponding to each candidate area in a preset candidate area set that the target device is located;

[0008] Selecting a target region from a plurality of candidate regions according to the first probability, wherein the target region includes at least one target sub-region;

[0009] Calling wireless signal connection information of the target area, the wireless signal connection information carrying data generated by connecting to wireless resources in a historical period;

[0010] Calculating a second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information;

[0011] The location of the target device is acquired according to the first probability and the second probability.

[0012] Accordingly, an embodiment of the present application further provides a data processing device, including:

[0013] A collection module is used to collect the latitude and longitude information of the target device and the wireless signal access information;

[0014] A first calculation module, configured to calculate a first probability corresponding to each candidate area in a preset candidate area set that the target device is located in according to the latitude and longitude information and the wireless signal access information;

[0015] A selection module, configured to select a target region from a plurality of candidate regions according to the first probability, wherein the target region includes at least one target sub-region;

[0016] A calling module, used to call the wireless signal connection information of the target area, wherein the wireless signal connection information carries data generated by connecting to wireless resources in a historical period;

[0017] A second calculation module, configured to calculate a second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information;

[0018] An acquisition module is used to acquire the location of the target device according to the first probability and the second probability.

[0019] Optionally, in some embodiments of the present application, the first calculation module includes:

[0020] A first acquisition unit, used to acquire a preset decision tree model and a sequence neural network classification model;

[0021] A first calculation unit, configured to calculate the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set of the target device according to the decision tree model and the longitude and latitude information;

[0022] A second calculation unit is used to calculate the wireless signal positioning probability corresponding to each candidate area in the preset candidate area set that the target device reaches based on the sequence neural network classification model and the wireless signal access information;

[0023] The fusion unit is used to fuse the longitude and latitude positioning probability and the wireless signal positioning probability to obtain a first probability corresponding to each candidate area in the preset candidate area set that the target device is located.

[0024] Optionally, in some embodiments of the present application, the first computing unit includes:

[0025] An output subunit, used to input the latitude and longitude information into each tree in the decision tree model, and output the score value corresponding to each tree;

[0026] The summing unit is used to sum the multiple score values ​​to obtain the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set that the target device reaches.

[0027] Optionally, in some embodiments of the present application, the output subunit is specifically used for:

[0028] Extracting the feature vector of the longitude and latitude information through the decision tree model;

[0029] Inputting the feature vector into each tree in the decision tree model to obtain a leaf node corresponding to each tree, wherein each leaf node corresponds to a constraint condition;

[0030] Detecting whether the leaf node meets the corresponding constraint condition;

[0031] Based on whether the leaf nodes meet the corresponding constraints, output the score value corresponding to each tree.

[0032] Optionally, in some embodiments of the present application, the second computing unit is specifically configured to:

[0033] Extracting the physical address of the target device and the wireless signal access strength value from the wireless signal access information;

[0034] Constructing a wireless signal access feature corresponding to the wireless signal access information based on the physical address of the target device and the wireless signal access strength value;

[0035] The wireless signal positioning probability of the target device reaching each candidate area in the preset candidate area set is calculated by using the sequential neural network classification model and the wireless signal access characteristics.

[0036] Optionally, in some embodiments of the present application, the selection module is specifically used to:

[0037] Determine the candidate area whose longitude and latitude positioning probability is greater than a first preset value as the target area, or;

[0038] The candidate area where the wireless signal positioning probability is greater than a first preset value is determined as the target area.

[0039] Optionally, in some embodiments of the present application, the second calculation module is specifically used to:

[0040] Obtain a preset sequence neural network matching model;

[0041] Extracting the physical address of the target device from the wireless signal access information, and constructing a physical address sequence corresponding to the physical address;

[0042] Acquire the connection frequency and wireless signal strength value of each wireless resource from the wireless signal connection information;

[0043] constructing a wireless signal connection sequence corresponding to the wireless signal connection information based on the connection frequency of the wireless resource and the wireless signal strength value;

[0044] splicing the physical address sequence and the wireless signal connection sequence to obtain a spliced ​​sequence;

[0045] The spliced ​​sequence is input into the sequence neural network matching model, and a second probability corresponding to the target device reaching each target sub-region is output.

[0046] Optionally, in some embodiments of the present application, a retention module is further included, and the retention module is specifically used to:

[0047] Determine the target sub-region whose second probability is greater than the second preset value as a candidate sub-region, and retain the candidate sub-region;

[0048] The acquisition module is specifically configured to acquire the location of the target device according to the first probability and a second probability corresponding to the candidate sub-area.

[0049] Optionally, in some embodiments of the present application, the acquisition module is specifically used to:

[0050] Calculate the posterior probability of the second probability based on Bayes' rule;

[0051] Calculate the product of the posterior probability and the first probability to obtain a joint probability;

[0052] The target sub-region corresponding to the largest joint probability is determined as the location of the target device.

[0053] After collecting the longitude and latitude information and wireless signal access information of the target device, the embodiment of the present application calculates the first probability corresponding to the target device being located in each candidate area in the preset candidate area set according to the longitude and latitude information and the wireless signal access information, and then selects the target area from multiple candidate areas according to the first probability, the target area includes at least one target sub-area, and then calls the wireless signal connection information of the target area, the wireless signal connection information carries the data generated by connecting to the wireless resource in the historical period, and then, based on the wireless signal access information and the wireless signal connection information, calculates the second probability corresponding to the target device reaching each target sub-area, and finally, obtains the position of the target device according to the first probability and the second probability. Therefore, this solution can improve the accuracy of device positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0055] Figure 1a It is a scenario schematic diagram of the data processing method provided in the embodiment of the present application;

[0056] Figure 1b It is a flowchart of a data processing method provided in an embodiment of the present application;

[0057] Figure 1c This is a schematic diagram of user store visits provided by this application;

[0058] Figure 1d This is a schematic diagram of the structure of the BERT model;

[0059] Figure 1e It is a schematic diagram of the structure of the BERT model provided in this application;

[0060] Figure 2 is another flow chart of the data processing method provided in the embodiment of the present application;

[0061] Figure 3a is a structural schematic diagram of a data processing device provided in an embodiment of the present application;

[0062] Figure 3b is another structural schematic diagram of a data processing device provided in an embodiment of the present application;

[0063] Figure 4 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0065] Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0066] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0067] Machine Learning (ML) is a multi-disciplinary subject that involves probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence.

[0068] Deep learning is the core part of machine learning, which usually includes artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and teaching learning. Among them, deep learning is a new research direction in the field of machine learning. That is, deep learning is a method in machine learning based on characterizing and learning data. Observations (such as an image) can be represented in a variety of ways, such as a vector of intensity values ​​for each pixel, or more abstractly represented as a series of edges, areas of specific shapes, etc. Using certain specific representation methods makes it easier to learn tasks from examples (for example, face recognition or facial expression recognition). The benefit of deep learning is to replace manual feature acquisition with efficient algorithms for unsupervised or semi-supervised feature learning and hierarchical feature extraction.

[0069] With the research and advancement of artificial intelligence technology, artificial intelligence technology has been studied and applied in many fields, such as common smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, unmanned driving, automatic driving, drones, robots, smart medical care, smart customer service, etc. I believe that with the development of technology, artificial intelligence technology will be applied in more fields and play an increasingly important role.

[0070] The solution provided in the embodiments of the present application involves technologies such as deep learning of artificial intelligence, which is now illustrated by the following embodiments.

[0071] Embodiments of the present application provide a data processing method, device, electronic device, and storage medium.

[0072] Among them, the data processing device can be specifically integrated in a terminal or a server. The terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc., but is not limited to this. The terminal and the server can be directly or indirectly connected through wired or wireless communication. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. This application does not limit this.

[0073] For example, see Figure 1a The present application provides a data processing system, including a mobile phone 10 and a server 20, wherein the data processing device is integrated in the server 20. Specifically, after a user enters a shopping mall, he opens the positioning service through the mobile phone 10, and the server 20 collects the latitude and longitude information and wireless signal access information of the target device. Then, the server 20 calculates the first probability corresponding to each candidate area (i.e., business district) in the preset candidate area set (i.e., shopping mall) based on the longitude and latitude information and the wireless signal access information. Then, the server 20 selects a target area from multiple candidate areas based on the first probability, and the target area includes at least one target sub-area (i.e., shop). Then, the server 20 calls the wireless signal connection information of the target area, and the wireless signal connection information carries the data generated by connecting to the wireless resource in the historical period. Then, the server 20 calculates the second probability corresponding to the mobile phone 10 reaching each target sub-area based on the wireless signal access information and the wireless signal connection information. Finally, the server 20 locates the position of the mobile phone 10 based on the first probability and the second probability.

[0074] The data processing solution provided by the present application utilizes the latitude and longitude information of the target device and the wireless signal access information to recall the target area, and utilizes the wireless signal access information and the wireless signal connection information to calculate the probability of the target device reaching each target sub-area. That is, the present solution mainly relies on wireless signals, which can avoid the problem of inaccurate positioning accuracy caused by large errors in indoor longitude and latitude, and when the range of the candidate area is large, there is no need to perform multiple repeated calculations, thereby reducing the amount of calculation during positioning while improving the accuracy of positioning.

[0075] It should be noted that the order of description of the following embodiments is not intended to limit the priority order of the embodiments.

[0076] A data processing method includes: collecting longitude and latitude information and wireless signal access information of a target device, calculating a first probability corresponding to each candidate area in a preset candidate area set that the target device is located according to the longitude and latitude information and the wireless signal access information, selecting a target area from multiple candidate areas according to the first probability, calling wireless signal connection information of the target area, calculating a second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information, and acquiring the position of the target device according to the first probability and the second probability.

[0077] See also Figure 1b , Figure 1b The following is a flow chart of a data processing method provided in an embodiment of the present application. The specific flow of the data processing method may be as follows:

[0078] 101. Collect the latitude and longitude information and wireless signal access information of the target device.

[0079] The target device can be any computing device that can receive base station signals and can communicate with the server. For example, the target device can be a cellular phone, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA), a handheld device with wireless communication capabilities, a computing device or other processing device connected to a wireless modem, a vehicle-mounted device, a wearable device, a terminal device in a future 5G network, or a terminal device in a future evolved PLMN (Public Land Mobile Network), etc.

[0080] Among them, the latitude and longitude information of the target device can be the positioning information generated by the global positioning system. The global positioning system (Global Positioning System, GPS) is a high-precision radio navigation positioning system based on artificial earth satellites. It can provide accurate geographical location, vehicle speed and precise time information anywhere in the world and in near-Earth space. The target device can be equipped with a GPS receiver, and the GPS receiver is used to determine the distance from the target device to the GPS satellite, and the three-dimensional position, three-dimensional motion speed, clock error and other parameters of the target device are calculated based on the information such as the position of the satellite in space at the moment of observation given by the satellite ephemeris.

[0081] Wireless signal access information is WiFi signal access information. WiFi signal is essentially a wireless point. Radio refers to electromagnetic waves propagating in all free spaces (including air and vacuum). It is a limited frequency band with an upper limit frequency of 3THz (terahertz) and a lower limit frequency of less uniform. Common ones are 3KHz~3THz (ITU-International Telecommunication Union regulations), 9KHz~3THz, and 10KHz~3THz.

[0082] 102. Calculate a first probability corresponding to each candidate area in a preset candidate area set that the target device is located according to the latitude and longitude information and the wireless signal access information.

[0083] The preset candidate area set includes at least one candidate area, and the candidate area may be a pre-set area, such as a business district in a shopping mall or a floor in a shopping mall, and is specifically set according to actual conditions. Further, the latitude and longitude information is processed using a preset decision tree model, and the wireless signal access information is processed using a preset neural network model, and the first probability corresponding to each candidate area in the preset candidate area set is calculated. That is, optionally, in some embodiments, the step of "calculating the first probability corresponding to each candidate area in the preset candidate area set for the target device based on the latitude and longitude information and the wireless signal access information" may specifically include:

[0084] (11) Obtaining a preset decision tree model and a sequence neural network classification model;

[0085] (12) Calculating the longitude and latitude positioning probability of the target device reaching each candidate area in the preset candidate area set based on the decision tree model and the longitude and latitude information;

[0086] (13) Based on the sequential neural network classification model and the wireless signal access information, the wireless signal positioning probability corresponding to each candidate area in the preset candidate area set is calculated;

[0087] (14) The longitude and latitude positioning probability and the wireless signal positioning probability are integrated to obtain a first probability corresponding to each candidate area in the preset candidate area set where the target device is located.

[0088] Among them, the decision tree model can be an XGBoost model, which is a type of gradient boosting tree model. It also generates models serially and takes the sum of all models as output. The XGBoost model performs a second-order Taylor expansion on the loss function, optimizes the loss function using the second-order derivative information of the loss function, and chooses whether to split the node based on the loss function. At the same time, XGBoost adds regularization, learning rate, column sampling, and approximate optimal split points to prevent overfitting.

[0089] It should be noted that decision tree is a non-parametric supervised learning method, which is mainly used for classification and regression. The purpose of decision tree is to construct a model that can predict the value of the target variable through decision rules from the characteristic attributes of sample data.

[0090] Decision trees often use a top-down design method. Each time an iteration cycle is completed, a feature attribute is selected for bifurcation until no more bifurcation is possible. Therefore, in the process of building a decision tree, the key is to select the best bifurcation feature attribute (which can quickly classify and make the depth of the decision tree small). This "optimality" can be measured by impurity. If there is only one classification result in a data set, the set is the purest, that is, the consistency is good; conversely, if there are many classifications, it is impure, that is, the consistency is not good. There are many indicators that can quantitatively measure this impurity, the most commonly used ones are entropy, Gini Index and classification error.

[0091] The XGBoost model is a strong learner integrated by multiple weak learners. The weak learners forming a strong learner involves a concept: ensemble learning. Ensemble learning refers to the combination of several weak classifiers to produce a strong classifier. Weak classifiers refer to those whose classification accuracy is only slightly better than random guessing (error rate <50%), while the accuracy of strong learners is higher than that of weak classifiers. XGBoost mainly uses feature dimensions to divide the sample space, adds trees by continuously splitting features, and finally obtains a model containing K trees through a greedy algorithm and hyperparameters that adjust the maximum depth of the tree, and uses the model to predict the sample. Specifically, the longitude and latitude information is input into each tree in the decision tree model, and then the score values ​​output by each tree are summed to obtain the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set for the positioning device. That is, optionally, in some embodiments, the step of "calculating the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set for the target device according to the decision tree model and the longitude and latitude information" can specifically include:

[0092] (21) Input the longitude and latitude information into each tree in the decision tree model, and output the score value corresponding to each tree;

[0093] (22) Multiple score values ​​are summed to obtain the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set.

[0094] Taking multiple weak learners as CART (Classification And Regression Tree) as an example, the longitude and latitude information is input into the XGBoost model, and the feature vector of the longitude and latitude information is extracted by the XGBoost model. The feature vector is input into the multiple weak learners, and each weak learner can perform feature splitting on the feature vector to obtain the leaf node of the decision tree where the weak learner is located. Since each weak learner corresponds to a constraint condition, where the constraint condition is different in different scenarios, under the regression tree, the constraint condition of the weak learner is: the square error minimization criterion, where the square error is used to represent the prediction error of the regression tree for the training data, that is, when the prediction error of the weak learner is the smallest, it is determined that the weak learner meets the constraint condition; under the classification tree, the constraint condition of the weak learner is: the Gini index minimization criterion, where the Gini index is an inequality measure that can be used to measure any uneven distribution, and the Gini index is a number between 0 and 1, 0-completely equal, 1-completely unequal, the more chaotic the categories contained in the population, the larger the Gini index (very similar to the concept of entropy). In the CART algorithm, Gini impurity indicates the possibility that a randomly selected sample is misclassified in a subset. Gini impurity is the probability of the sample being selected multiplied by the probability that it is misclassified. When all samples in a node are of the same class, Gini impurity is zero. That is, when the Gini index of a weak learner is the smallest, it is determined that the weak learner meets the constraint condition.

[0095] For leaf nodes that meet the constraint conditions, the weak learner outputs a leaf node score greater than 0. For leaf nodes that do not meet the constraint conditions, the weak learner outputs a leaf node score less than 0. The above process is repeated, and multiple weak learners will output multiple leaf node scores. By summing the multiple leaf node scores, the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set of the target device is obtained. That is, optionally, in some embodiments, the step of "inputting the longitude and latitude information into each tree in the decision tree model and outputting the score value corresponding to each tree" may specifically include:

[0096] (31) Extracting feature vectors of longitude and latitude information through a decision tree model;

[0097] (32) Input the feature vector into each tree in the decision tree model to obtain the leaf node corresponding to each tree;

[0098] (33) Check whether the leaf node meets the corresponding constraint conditions;

[0099] (34) Based on whether the leaf nodes meet the corresponding constraints, output the score value corresponding to each tree.

[0100] Among them, the constraints can be flexibly set according to the actual situation. In this application, it is necessary to predict the longitude and latitude positioning probability corresponding to each candidate area. Taking into account that the candidate areas correspond to different longitudes and latitudes, the candidate areas at different altitudes are first distinguished according to the longitude, and then the positions of the candidate areas are distinguished by the latitude. Finally, based on the longitude and latitude information of the target device, the longitude and longitude positioning probability corresponding to each candidate area in the preset candidate area set is calculated one by one.

[0101] It should be noted that the XGBoost model can be pre-trained in advance. Taking the indoor positioning in shopping mall A as an example for specific explanation, first, the latitude and longitude information of users who entered shopping mall A in the past month is collected. For the convenience of description, it is referred to as samples below, where the sample is marked with the area where the user is located. The area includes stores within the area and stores not within the area. It should be noted that stores within the area can be understood as being within the area and entering the store. Similarly, stores not within the area can be understood as being within the area but not entering the store. Please refer to Figure 1c In the figure, circle a represents store visits within the area, and rectangle b represents non-store visits within the area. Then, the samples are input into the XGBoost model to predict the user's store visits. Finally, based on the user's label and the user's prediction results, the network parameters of the XGBoost model are adjusted until the learning is completed, and a trained XGBoost model is obtained. The trained XGBoost model has the ability to predict the user's store visits based on latitude and longitude information.

[0102] Further, in the present application, the sequence neural network can be a bidirectional encoder model based on transformers (BiDirectional Encoder Rpresentation From Transformers, BERT). The network architecture of BERT uses a multi-layer Transformers-encoder structure. Transformers-encoder is a neural network structure open sourced by Google for solving sequence problems. The input encoding vector (length is 512) of the BERT model is the unit sum of three embedding features. These three word embedding features are: word embedding, position embedding and segmentation embedding. Please refer to Figure 1d, BERT's input is the superposition of token, segmentation and position embeddings, that is, the embedding of each word is the superposition of three embeddings. For token embeddings, it is implemented through a word list; while positional embeddings are used to represent the position information of words in a sentence; segmentation embeddings are for the sentence as a whole, that is, used to distinguish two sentences, such as whether B is the context of A. Two sentences, such as whether B is the context of A, BERT is a bidirectional language model, which needs to consider both the context of words. To achieve this goal, the model adopts a simple method: randomly masking some input tokens, and then training the model to correctly predict those masked tokens. Specifically, the model randomly selects 15% of the words in the corpus, and then 80% of them will be replaced by [Mask] masks to obtain masked text strings, 10% of which will be randomly replaced by another word, and the remaining 10% will remain unchanged. The model is then required to correctly predict the selected words. The reason for doing these processing is that [Mask] only appears during training. If all selected words become [Mask], the trained model may be targeted at [Mask], which is inconsistent with reality. Therefore, there is a probability that [Mask] will be randomly replaced, and this probability is 10%×15%=1.5%, which basically does not damage the model's language understanding ability. In order to predict correctly, the model needs to maintain the distributed context representation of each input token, thereby achieving the purpose of feature extraction from the context. The previous masked language model is trained at the word level, and many natural language recognition tasks are at the sentence level. This requires the language model to understand the relationship between sentences, which is helpful for downstream sentence relationship judgment tasks, and the next sentence prediction in the BERT model is for this purpose.Its principle is also very simple, which is to pre-train a binary next sentence prediction task. Specifically, sentences A and B are selected as training samples: A's next sentence has a 50% chance of being B, and the other 50% chance that B is randomly taken from the corpus. The model needs to correctly predict whether B is the next sentence of A. For other tasks, corresponding predictions can be made based on the output information of the BERT model. In this application, a wireless signal access feature of the wireless signal access information can be constructed, that is, a feature sequence (wireless signal access feature) for characterizing the wireless signal is constructed using the physical address of the wireless signal and the wireless signal access strength value. Then, the wireless signal access feature is input into the BERT model to calculate the wireless signal positioning probability corresponding to the target device reaching each candidate area in the preset candidate area set, that is, optionally, in some embodiments, the step of "calculating the wireless signal positioning probability corresponding to the target device reaching each candidate area in the preset candidate area set based on the sequence neural network classification model and the wireless signal access information" can specifically include:

[0103] (41) extracting the physical address of the target device and the wireless signal access strength value from the wireless signal access information;

[0104] (42) constructing a wireless signal access feature corresponding to the wireless signal access information based on the physical address of the target device and the wireless signal access strength value;

[0105] (43) The wireless signal positioning probability of the target device reaching each candidate area in the preset candidate area set is calculated through the sequential neural network classification model and the wireless signal access characteristics.

[0106] It should be noted that the wireless signal access strength value, also known as the received signal strength (RSS), is used to measure the strength of the accessed wireless signal. The physical address, also known as the media access control address (MAC), is also called the local area network address (LAN Address), Ethernet address (Ethernet Address) or physical address (Physical Address). It is an address used to confirm the location of a network device. Here, we need to extend a concept, the basic service set (Basic Service Set) of the infrastructure type. The infrastructure basic service set (BSS) is defined within the transmission range of the access point. The access point is responsible for all communications of the infrastructure network, including communications of all mobile nodes in the same service area. The BSS can be used to provide WIFI services. The BSSID is the virtual MAC address of the device. This MAC can be the same as the actual MAC of the device, or the last digit can be changed based on the actual MAC. Different manufacturers have different implementation methods. This application constructs wireless signal access features corresponding to wireless signal access information based on BSSID and RSS. The specific forms are as follows: Bssid1:rssi1; Bssid2:rssi2; Bssid3:rssi3; ... BssidN:rssin. During prediction, the wireless signal access features are input into the BERT model to output the wireless signal positioning probability corresponding to the target device reaching each candidate area in the preset candidate area set.

[0107] The BERT model can be pre-trained. For details, please refer to Figure 1e , which is different from the input of the conventional BERT model, in this application, the BERT model does not make context predictions, therefore, the masked text string B is set to a sequence of all 0s, and the masked text string A is a mac address sequence and its strength value. For the specific form, please refer to the previous embodiment, wherein each group of masked text strings A is marked with the area where the user is located, and the area includes stores that have been to the area and stores that have not been to the area. After the masked text strings A and B are input into the BERT model, the first bit C of the output sequence is taken as the input of the classification layer, thereby predicting the wireless signal positioning probability corresponding to each candidate area in the candidate area set.

[0108] In summary, the longitude and latitude positioning probability and the wireless signal positioning probability are obtained, and the candidate areas with longitude and latitude positioning probability greater than 0.5 or the candidate areas with wireless signal positioning probability greater than 0.5 are retained. It should be noted that when the longitude and latitude positioning probability and the wireless signal positioning probability of the candidate area are both greater than 0.5, the larger probability value is taken as its first probability. For example, if the longitude and latitude positioning probability is less than 0.6 and the wireless signal positioning probability is 0.7, then the first probability of the candidate area is 0.7, and then step 103 is executed.

[0109] 103. Select a target area from multiple candidate areas according to the first probability.

[0110] Among them, the target area includes at least one target sub-area. After obtaining the first probability, the candidate area whose longitude and latitude positioning probability is greater than the first preset value is determined as the target area, or; the candidate area whose wireless signal positioning probability is greater than the first preset value is determined as the target area. Of course, the candidate area whose longitude and latitude positioning probability is greater than the first preset value and whose wireless signal positioning probability is greater than the first preset value can also be determined as the target area, and the selection is made according to the actual situation.

[0111] 104. Call the wireless signal connection information of the target area.

[0112] Among them, the wireless signal connection information carries the data generated by connecting to wireless resources during a historical period. For example, the wireless signal connection information carries the number of users who connected to the wireless resources of store A in the past month, as well as the wireless signal access strength of each user connected to the wireless resources.

[0113] 105. Calculate a second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information.

[0114] For example, specifically, the wireless signal access information and the wireless signal connection information can be input into the sequence neural network matching model to calculate the second probability corresponding to the target device reaching each target sub-area, wherein the sequence neural network matching model can also be a BERT model. Unlike the sequence neural network classification model, in the sequence neural network matching model, the masked text string A is the MAC address sequence received by the user, and the masked text string B is an aggregate sequence composed of the MAC addresses of historical visitors to the target sub-area (i.e., the store) and the signal access strength values ​​when connecting to the wireless resource. For example, the MAC address of the historical visitor to the store is 08:00:20:0A:8C:6D, and the signal access strength value when connecting to the wireless resource is -100db, then the corresponding aggregate sequence is 0800200A8C6D100.

[0115] In order to reduce the amount of calculation, the MAC addresses of the top 10 visitors with the highest number of visits (i.e., connection frequency) and the signal access strength values ​​when connecting to the wireless resource can be taken, and then the MAC address sequence received by the user is spliced ​​with the aggregate sequence, and the spliced ​​sequence is then input into the sequence neural network matching model to output the second probability corresponding to the target device reaching each target sub-area, that is, optionally, in some embodiments, the step of "calculating the second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information" can specifically include:

[0116] (51) obtaining a preset sequence neural network matching model;

[0117] (52) extracting the physical address of the target device from the wireless signal access information, and constructing a physical address sequence corresponding to the physical address;

[0118] (53) obtaining the connection frequency and wireless signal strength value of each wireless resource from the wireless signal connection information;

[0119] (54) constructing a wireless signal connection sequence corresponding to the wireless signal connection information based on the connection frequency of the wireless resource and the wireless signal strength value;

[0120] (55) splicing the physical address sequence and the wireless signal connection sequence to obtain a spliced ​​sequence;

[0121] (56) The concatenated sequence is input into a sequence neural network matching model, and a second probability corresponding to the target device reaching each target sub-region is output.

[0122] Furthermore, after obtaining the second probability, the target sub-region whose second probability is greater than the second preset value can be determined as a candidate sub-region, and the candidate sub-region is retained, and then step 105 is executed. It should be noted that the first preset value and the second preset value may be equal or unequal, and are set according to actual conditions.

[0123] 106. Obtain a location of the target device according to the first probability and the second probability.

[0124] Specifically, the product between the posterior probability of the second probability and the first probability is calculated according to the Bayesian rule, and then the target sub-area corresponding to the largest joint probability is determined as the location of the target device. It should be noted that the Bayesian rule is also called the Bayesian theorem. The Bayesian rule is a standard method for correcting the subjective judgment (i.e., prior probability) of the probability distribution based on the observed phenomena in the application of probability statistics. The so-called Bayesian rule means that when the analysis sample is large enough to approach the total number, the probability of an event in the sample will be close to the probability of the event in the total number. Usually, the probability of event A under the condition of event B (occurrence) is different from the probability of event B under the condition of event A; however, there is a definite relationship between the two, and the Bayesian rule is a statement of this relationship, such as P(A|B) and P(B|A). According to the multiplication rule, it can be immediately derived that: P(A∩B)=P(A)*P(B|A)=P(B)*P(A|B), where P(A|B) is the conditional probability of A after it is known that B occurs, and is also called the posterior probability of A because it is obtained from the value of B. It can be seen that the posterior probability of the second probability is the probability that the target device belongs to the target sub-area in the candidate area, that is, P=P(first probability, second probability)=P(first probability)x P(second probability|first probability). Optionally, in some embodiments, the location of the target device can be obtained based on the first probability and the second probability corresponding to the candidate sub-area.

[0125] After collecting the longitude and latitude information and wireless signal access information of the target device, the embodiment of the present application calculates the first probability corresponding to the target device being located in each candidate area in the preset candidate area set according to the longitude and latitude information and the wireless signal access information, then selects the target area from multiple candidate areas according to the first probability, then calls the wireless signal connection information of the target area, and then, based on the wireless signal access information and the wireless signal connection information, calculates the second probability corresponding to the target device reaching each target sub-area, and finally, obtains the position of the target device according to the first probability and the second probability. The data processing scheme provided by the present application uses the longitude and latitude information and the wireless signal access information of the target device to recall the target area, and uses the wireless signal access information and the wireless signal connection information to calculate the probability of the target device reaching each target sub-area, that is, the scheme uses the wireless signal access information and the longitude and latitude information to recall the candidate area, and then obtains the position of the target device in the candidate area according to the wireless signal access information and the wireless signal connection information, which can avoid the problem of inaccurate positioning accuracy caused by large errors in indoor longitude and latitude, and when the range of the candidate area is large, there is no need to perform multiple repeated calculations, thereby reducing the amount of calculation during positioning and improving the accuracy of positioning.

[0126] The method described in the embodiment will be further described in detail below with examples.

[0127] In this embodiment, the data processing device is specifically integrated into a server as an example for description.

[0128] See also Figure 2 , a data processing method, the specific process can be as follows:

[0129] 201. The server collects the latitude and longitude information and wireless signal access information of the target device.

[0130] The server obtains the latitude and longitude information and wireless signal access information reported by the target device from the database.

[0131] 202. The server calculates a first probability corresponding to each candidate area in a preset candidate area set that the target device is located according to the latitude and longitude information and the wireless signal access information.

[0132] Among them, the preset candidate area set includes at least one candidate area, and the candidate area can be pre-set, such as a business district in a shopping mall or a floor in a shopping mall. Furthermore, the server can use a preset decision tree model to process the longitude and latitude information, and use a preset neural network model to process the wireless signal access information to estimate the first probability corresponding to each candidate area in the preset candidate area set that the target device is located.

[0133] 203. The server selects a target area from multiple candidate areas according to the first probability.

[0134] Among them, the target area includes at least one target sub-area. After obtaining the first probability, the candidate area whose longitude and latitude positioning probability is greater than the first preset value is determined as the target area, or; the candidate area whose wireless signal positioning probability is greater than the first preset value is determined as the target area. Of course, the candidate area whose longitude and latitude positioning probability is greater than the first preset value and whose wireless signal positioning probability is greater than the first preset value can also be determined as the target area, and the selection is made according to the actual situation.

[0135] 204. The server calls the wireless signal connection information of the target area.

[0136] Among them, the wireless signal connection information carries the data generated by connecting to wireless resources during a historical period. For example, the wireless signal connection information carries the number of users who connected to the wireless resources of store A in the past month, as well as the wireless signal access strength of each user connected to the wireless resources.

[0137] 205. The server calculates a second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information.

[0138] For example, specifically, the server can input the wireless signal access information and the wireless signal connection information into the sequential neural network matching model to calculate the second probability corresponding to the target device reaching each target sub-area, wherein the sequential neural network matching model can also be a BERT model. Unlike the sequential neural network classification model, in the sequential neural network matching model, the masked text string A is a MAC address sequence received by the user, and the masked text string B is an aggregate sequence composed of the MAC addresses of historical visitors to the target sub-area (i.e., the store) and the signal access strength values ​​when connecting to the wireless resource.

[0139] 206. The server obtains the location of the target device according to the first probability and the second probability.

[0140] Specifically, the server may calculate the product of the posterior probability of the second probability and the first probability according to the Bayesian rule, and then determine the target sub-area corresponding to the maximum joint probability as the location of the target device.

[0141] In order to facilitate further understanding of the data processing method of the present application, the following is a specific description taking a user visiting a shopping mall S as an example. The shopping mall S includes business districts A, B and C. Business district A includes at least one store, business district B includes at least one store, and business district C includes at least one store. When the user visits the shopping mall S, the server can obtain the latitude and longitude information and wireless signal access information of the user's mobile phone (i.e., the target device). The server calculates the first probability corresponding to the mobile phone being located in each business district in the shopping mall S based on the latitude and longitude information and the wireless signal access information. Then, the server selects the target business district from the business districts A, B and C based on the first probability. For example, the server selects the target business district based on the latitude and longitude information and the wireless signal access information. The server calculates the probability that the mobile phone is located in business district A as 0.8, the probability that the mobile phone is located in business district B as 0.6, and the probability that the mobile phone is located in business district A as 0.5. The server can determine business district A as the target business district. Then, the server calls the wireless signal connection information of business district A, and the wireless signal connection information carries the data generated by connecting to the wireless resource in the historical period. Then, the server calculates the second probability corresponding to the mobile phone reaching each store in business district A based on the wireless signal access information and the wireless signal connection information. Finally, the server obtains the location of the mobile phone according to the first probability and the second probability. For example, the server can use the following formula to calculate the probability that the user visits the store, that is, the probability that the mobile phone is located in the store P = P (first probability) x P (second probability | first probability), sort according to the visit probability calculated as above, retain the stores corresponding to the probability P greater than 0.5, and select the store with the highest probability from the retained stores to determine as the store that the user will visit. If the probability P of the mobile phone being located in the store is less than 0.5, then output "not visited".

[0142] After the server of the embodiment of the present application collects the latitude and longitude information and the wireless signal access information of the target device, the server calculates the first probability corresponding to the target device being located in each candidate area in the preset candidate area set based on the latitude and longitude information and the wireless signal access information. Then, the server selects the target area from multiple candidate areas based on the first probability. Next, the server calls the wireless signal connection information of the target area. Then, the server calculates the second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information. Finally, the server obtains the location of the target device based on the first probability and the second probability. The data processing scheme provided by the present application utilizes the latitude and longitude information and the wireless signal access information of the target device to recall the target area, and utilizes the wireless signal access information and the wireless signal connection information to calculate the probability of the target device reaching each target sub-area, that is, the present scheme utilizes the wireless signal access information and the longitude and latitude information to recall the candidate area, and then obtains the position of the target device in the candidate area based on the wireless signal access information and the wireless signal connection information, thereby avoiding the problem of inaccurate positioning accuracy caused by large errors in indoor longitude and latitude, and when the candidate area is large, there is no need to perform multiple repeated calculations, thereby improving the accuracy of positioning while reducing the amount of calculation during positioning.

[0143] In order to facilitate better implementation of the data processing method of the embodiment of the present application, the embodiment of the present application also provides a data processing device based on the above (hereinafter referred to as the processing device). The meaning of the terms is the same as in the above data processing method, and the specific implementation details can refer to the description in the method embodiment.

[0144] See also Figure 3a , Figure 3a A schematic diagram of the structure of a data processing device provided in an embodiment of the present application, wherein the interactive device may include a collection module 301, a first calculation module 302, a selection module 303, a calling module 304, a second calculation module 305 and an acquisition module 306, which may be specifically as follows:

[0145] The collection module 301 is used to collect the latitude and longitude information and wireless signal access information of the target device.

[0146] The collection module 301 obtains the latitude and longitude information and wireless signal access information reported by the target device from the database.

[0147] The first calculation module 302 is used to calculate the first probability corresponding to each candidate area in the preset candidate area set that the target device is located according to the longitude and latitude information and the wireless signal access information.

[0148] Among them, the preset candidate area set includes at least one candidate area, and the first calculation module 302 can use a preset decision tree model to process the longitude and latitude information, and use a preset neural network model to process the wireless signal access information to estimate the first probability corresponding to each candidate area in the preset candidate area set that the target device is located.

[0149] Optionally, in some embodiments, the first calculation module 302 may specifically include:

[0150] A first acquisition unit, used to acquire a preset decision tree model and a sequence neural network classification model;

[0151] A first calculation unit, configured to calculate the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set according to the decision tree model and the longitude and latitude information;

[0152] A second calculation unit is used to calculate the wireless signal positioning probability corresponding to each candidate area in the preset candidate area set that the target device reaches based on the sequence neural network classification model and the wireless signal access information;

[0153] The fusion unit is used to fuse the longitude and latitude positioning probability and the wireless signal positioning probability to obtain a first probability corresponding to each candidate area in the preset candidate area set that the target device is located.

[0154] Optionally, in some embodiments, the first calculation unit may specifically include:

[0155] The output subunit is used to input the latitude and longitude information into each tree in the decision tree model and output the score value corresponding to each tree;

[0156] The summing unit is used to sum the multiple score values ​​to obtain the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set that the target device reaches.

[0157] Optionally, in some embodiments, the output subunit can be specifically used to: extract a feature vector of longitude and latitude information through a decision tree model; input the feature vector into each tree in the decision tree model to obtain a leaf node corresponding to each tree, wherein each leaf node corresponds to a constraint condition; detect whether the leaf node meets the corresponding constraint condition; and output the score value corresponding to each tree based on whether the leaf node meets the corresponding constraint condition.

[0158] Optionally, in some embodiments, the second calculation unit can be specifically used to: extract the physical address and wireless signal access strength value of the target device from the wireless signal access information; construct a wireless signal access feature corresponding to the wireless signal access information based on the physical address and wireless signal access strength value of the target device; and calculate the wireless signal positioning probability corresponding to the target device reaching each candidate area in the preset candidate area set through a sequential neural network classification model and the wireless signal access feature.

[0159] The selection module 303 is configured to select a target area from multiple candidate areas according to the first probability.

[0160] Among them, the target area includes at least one target sub-area. After obtaining the first probability, the selection module 303 determines the candidate area whose longitude and latitude positioning probability is greater than the first preset value as the target area, or; the selection module 303 determines the candidate area whose wireless signal positioning probability is greater than the first preset value as the target area.

[0161] Optionally, in some embodiments, the selection module 303 can be specifically used to: determine a candidate area whose longitude and latitude positioning probability is greater than a first preset value as a target area, or; determine a candidate area whose wireless signal positioning probability is greater than a first preset value as a target area.

[0162] The calling module 304 is used to call the wireless signal connection information of the target area.

[0163] The wireless signal connection information carries the data generated by connecting to wireless resources during the historical period.

[0164] The second calculation module 305 is used to calculate a second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information.

[0165] For example, specifically, the second calculation module 305 can input the wireless signal access information and the wireless signal connection information into the sequential neural network matching model to calculate the second probability corresponding to the target device reaching each target sub-area.

[0166] Optionally, in some embodiments, the second calculation module 305 can be specifically used to: obtain a preset sequence neural network matching model; extract the physical address of the target device from the wireless signal access information, and construct a physical address sequence corresponding to the physical address; obtain the connection frequency and the wireless signal strength value of each wireless resource from the wireless signal connection information; based on the connection frequency and the wireless signal strength value of the wireless resource, construct a wireless signal connection sequence corresponding to the wireless signal connection information; splice the physical address sequence and the wireless signal connection sequence to obtain a spliced ​​sequence; input the spliced ​​sequence into the sequence neural network matching model, and output the second probability corresponding to the target device reaching each target sub-area.

[0167] The acquisition module 306 is configured to acquire the location of the target device according to the first probability and the second probability.

[0168] Optionally, in some embodiments, the acquisition module 306 can be specifically used to: calculate the posterior probability of the second probability based on the Bayesian rule; calculate the product between the posterior probability and the first probability to obtain a joint probability; and determine the target sub-area corresponding to the largest joint probability as the location of the target device.

[0169] Optionally, in some embodiments, see Figure 3b The positioning device may further include a retaining module 307, and the retaining module 307 may be used to: determine the target sub-region whose second probability is greater than the second preset value as a candidate sub-region, and retain the candidate sub-region;

[0170] The acquisition module 306 may also be specifically configured to: acquire the location of the target device according to the first probability and the second probability corresponding to the candidate sub-region.

[0171] After the acquisition module 301 of the embodiment of the present application acquires the latitude and longitude information and the wireless signal access information of the target device, the first calculation module 302 calculates the first probability corresponding to the target device being located in each candidate area in the preset candidate area set based on the longitude and latitude information and the wireless signal access information. Then, the selection module 303 selects the target area from multiple candidate areas based on the first probability. Next, the calling module 304 calls the wireless signal connection information of the target area. Then, the second calculation module 305 calculates the second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information. Finally, the acquisition module 306 acquires the position of the target device based on the first probability and the second probability. The data processing scheme provided by the present application utilizes the latitude and longitude information and the wireless signal access information of the target device to recall the target area, and utilizes the wireless signal access information and the wireless signal connection information to calculate the probability of the target device reaching each target sub-area, that is, the present scheme utilizes the wireless signal access information and the longitude and latitude information to recall the candidate area, and then obtains the position of the target device in the candidate area based on the wireless signal access information and the wireless signal connection information, thereby avoiding the problem of inaccurate positioning accuracy caused by large errors in indoor longitude and latitude, and when the candidate area is large, there is no need to perform multiple repeated calculations, thereby improving the accuracy of positioning while reducing the amount of calculation during positioning.

[0172] In addition, the present application also provides an electronic device, such as Figure 4 As shown, it shows a schematic diagram of the structure of the electronic device involved in the embodiment of the present application, specifically:

[0173] The electronic device may include components such as a processor 401 with one or more processing cores, a memory 402 with one or more computer-readable storage media, a power supply 403, and an input unit 404. Those skilled in the art will appreciate that Figure 4 The electronic device structure shown in the figure does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0174] The processor 401 is the control center of the electronic device, and uses various interfaces and lines to connect various parts of the entire electronic device. By running or executing software programs and / or modules stored in the memory 402, and calling data stored in the memory 402, the processor 401 performs various functions of the electronic device and processes data, thereby performing overall detection of the electronic device. Optionally, the processor 401 may include one or more processing cores; preferably, the processor 401 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 401.

[0175] The memory 402 can be used to store software programs and modules. The processor 401 executes various functional applications and data processing by running the software programs and modules stored in the memory 402. The memory 402 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 402 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices. Accordingly, the memory 402 may also include a memory controller to provide the processor 401 with access to the memory 402.

[0176] The electronic device also includes a power supply 403 for supplying power to each component. Preferably, the power supply 403 can be logically connected to the processor 401 through a power management system, so as to manage charging, discharging, power consumption and other functions through the power management system. The power supply 403 can also include one or more DC or AC power supplies, recharging systems, power failure detection circuits, power converters or inverters, power status indicators and other arbitrary components.

[0177] The electronic device may further include an input unit 404, which may be used to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal input related to user settings and function control.

[0178] Although not shown, the electronic device may further include a display unit, etc., which will not be described in detail herein. Specifically in this embodiment, the processor 401 in the electronic device will load the executable files corresponding to the processes of one or more application programs into the memory 402 according to the following instructions, and the processor 401 will run the application programs stored in the memory 402, thereby realizing various functions, as follows:

[0179] Collect the latitude and longitude information and wireless signal access information of the target device, calculate the first probability corresponding to the target device being located in each candidate area in the preset candidate area set based on the latitude and longitude information and the wireless signal access information, select the target area from multiple candidate areas based on the first probability, call the wireless signal connection information of the target area, calculate the second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information, and obtain the position of the target device based on the first probability and the second probability.

[0180] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0181] After collecting the longitude and latitude information and wireless signal access information of the target device, the embodiment of the present application calculates the first probability corresponding to each candidate area in the preset candidate area set of the target device according to the longitude and latitude information and the wireless signal access information, and then selects the target area from multiple candidate areas according to the first probability, then calls the wireless signal connection information of the target area, and then, based on the wireless signal access information and the wireless signal connection information, calculates the second probability corresponding to the target device reaching each target sub-area, and finally, obtains the position of the target device according to the first probability and the second probability. The data processing scheme provided by the present application uses the longitude and latitude information and the wireless signal access information of the target device to recall the target area, and uses the wireless signal access information and the wireless signal connection information to calculate the probability of the target device reaching each target sub-area, that is, the scheme mainly relies on wireless signals, which can avoid the problem of inaccurate positioning accuracy caused by large errors in indoor longitude and latitude, and when the range of the candidate area is large, there is no need to perform multiple repeated calculations, thereby reducing the amount of calculation during positioning while improving the accuracy of positioning.

[0182] A person of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0183] To this end, an embodiment of the present application provides a storage medium in which a plurality of instructions are stored, and the instructions can be loaded by a processor to execute the steps in any data processing method provided in the embodiment of the present application. For example, the instructions can execute the following steps:

[0184] Collect the latitude and longitude information and wireless signal access information of the target device, calculate the first probability corresponding to the target device being located in each candidate area in the preset candidate area set based on the latitude and longitude information and the wireless signal access information, select the target area from multiple candidate areas based on the first probability, call the wireless signal connection information of the target area, calculate the second probability corresponding to the target device reaching each target sub-area based on the wireless signal access information and the wireless signal connection information, and obtain the position of the target device based on the first probability and the second probability.

[0185] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0186] The storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0187] Since the instructions stored in the storage medium can execute the steps in any data processing method provided in the embodiments of the present application, the beneficial effects that can be achieved by any data processing method provided in the embodiments of the present application can be achieved. Please refer to the previous embodiments for details and will not be repeated here.

[0188] The above is a detailed introduction to a data processing method, device, electronic device and storage medium provided in the embodiments of the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A data processing method, characterized in that: include: Collect the latitude and longitude information of the target device and the wireless signal access information; According to the longitude and latitude information and the wireless signal access information, calculating the first probability corresponding to the target device being located in each candidate area in the preset candidate area set, including: obtaining a preset decision tree model and a sequential neural network classification model; calculating the longitude and latitude positioning probability corresponding to the target device reaching each candidate area in the preset candidate area set according to the decision tree model and the longitude and latitude information; calculating the wireless signal positioning probability corresponding to the target device reaching each candidate area in the preset candidate area set based on the sequential neural network classification model and the wireless signal access information; fusing the longitude and latitude positioning probability and the wireless signal positioning probability to obtain the first probability corresponding to the target device being located in each candidate area in the preset candidate area set; Selecting a target region from a plurality of candidate regions according to the first probability, wherein the target region includes at least one target sub-region; Calling wireless signal connection information of the target area, the wireless signal connection information carrying data generated by connecting to wireless resources in a historical period; Based on the wireless signal access information and the wireless signal connection information, calculating the second probability corresponding to the target device arriving at each target sub-area, including: obtaining a preset sequence neural network matching model; extracting the physical address of the target device from the wireless signal access information, and constructing a physical address sequence corresponding to the physical address; obtaining the connection frequency and the wireless signal strength value of each wireless resource from the wireless signal connection information; constructing a wireless signal connection sequence corresponding to the wireless signal connection information based on the connection frequency and the wireless signal strength value of the wireless resource; splicing the physical address sequence and the wireless signal connection sequence to obtain a spliced ​​sequence; inputting the spliced ​​sequence into the sequence neural network matching model, and outputting the second probability corresponding to the target device arriving at each target sub-area; According to the first probability and the second probability, the location of the target device is obtained, including: calculating the posterior probability of the second probability based on the Bayesian rule; calculating the product between the posterior probability and the first probability to obtain a joint probability; and determining the target sub-area corresponding to the largest joint probability as the location of the target device.

2. The method according to claim 1, characterized in that The calculating, based on the decision tree model and the longitude and latitude information, the longitude and latitude positioning probability of the target device reaching each candidate area in the preset candidate area set includes: Input the latitude and longitude information into each tree in the decision tree model, and output the score value corresponding to each tree; The multiple score values ​​are summed to obtain the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set of the target device.

3. The method according to claim 2, characterized in that The step of inputting the latitude and longitude information into each tree in the decision tree model and outputting a score value corresponding to each tree includes: Extracting the feature vector of the longitude and latitude information through the decision tree model; Inputting the feature vector into each tree in the decision tree model to obtain a leaf node corresponding to each tree, wherein each leaf node corresponds to a constraint condition; Detecting whether the leaf node meets the corresponding constraint condition; Based on whether the leaf nodes meet the corresponding constraints, output the score value corresponding to each tree.

4. The method according to claim 1, characterized in that: The calculating, based on the sequence neural network classification model and the wireless signal access information, the wireless signal positioning probability corresponding to the target device reaching each candidate area in the preset candidate area set includes: Extracting the physical address of the target device and the wireless signal access strength value from the wireless signal access information; Constructing a wireless signal access feature corresponding to the wireless signal access information based on the physical address of the target device and the wireless signal access strength value; The wireless signal positioning probability of the target device reaching each candidate area in the preset candidate area set is calculated by using the sequential neural network classification model and the wireless signal access characteristics.

5. The method according to any one of claims 1 to 4, characterized in that: The selecting a target area from a plurality of candidate areas according to the first probability includes: Determine the candidate area whose longitude and latitude positioning probability is greater than a first preset value as the target area, or; The candidate area where the wireless signal positioning probability is greater than a first preset value is determined as the target area.

6. The method according to claim 1, characterized in that After outputting the second probability corresponding to the target device reaching each target sub-area, the method further includes: Determine the target sub-region where the second probability is greater than a second preset value as a candidate sub-region; The acquiring the position of the target device according to the first probability and the second probability includes: acquiring the position of the target device according to the first probability and a second probability corresponding to the candidate sub-area.

7. A data processing device, characterized in that: include: A collection module is used to collect the latitude and longitude information of the target device and the wireless signal access information; A first calculation module, configured to calculate a first probability corresponding to each candidate area in a preset candidate area set that the target device is located in according to the latitude and longitude information and the wireless signal access information; Among them, the first calculation module includes: an acquisition unit, a first calculation unit, a second calculation unit and a fusion unit; the acquisition unit is used to acquire a preset decision tree model and a sequential neural network classification model; the first calculation unit is used to calculate the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set of the target device according to the decision tree model and the longitude and latitude information; the second calculation unit is used to calculate the wireless signal positioning probability corresponding to each candidate area in the preset candidate area set of the target device based on the sequential neural network classification model and the wireless signal access information; the fusion unit is used to fuse the longitude and latitude positioning probability and the wireless signal positioning probability to obtain the first probability corresponding to each candidate area in the preset candidate area set of the target device; A selection module, configured to select a target region from a plurality of candidate regions according to the first probability, wherein the target region includes at least one target sub-region; A calling module, used to call the wireless signal connection information of the target area, wherein the wireless signal connection information carries data generated by connecting to wireless resources in a historical period; A second calculation module is used to calculate the second probability corresponding to the target device arriving at each target sub-area based on the wireless signal access information and the wireless signal connection information, including: obtaining a preset sequence neural network matching model; extracting the physical address of the target device from the wireless signal access information, and constructing a physical address sequence corresponding to the physical address; obtaining the connection frequency and the wireless signal strength value of each wireless resource from the wireless signal connection information; constructing a wireless signal connection sequence corresponding to the wireless signal connection information based on the connection frequency and the wireless signal strength value of the wireless resource; splicing the physical address sequence and the wireless signal connection sequence to obtain a spliced ​​sequence; inputting the spliced ​​sequence into the sequence neural network matching model, and outputting the second probability corresponding to the target device arriving at each target sub-area; An acquisition module is used to acquire the location of the target device according to the first probability and the second probability, including: calculating the posterior probability of the second probability based on the Bayesian rule; calculating the product between the posterior probability and the first probability to obtain a joint probability; and determining the target sub-area corresponding to the largest joint probability as the location of the target device.

8. The device according to claim 7, characterized in that The first computing unit comprises: An output subunit, used to input the latitude and longitude information into each tree in the decision tree model, and output the score value corresponding to each tree; The summing unit is used to sum the multiple score values ​​to obtain the longitude and latitude positioning probability corresponding to each candidate area in the preset candidate area set that the target device reaches.

9. The device according to claim 8, characterized in that The output subunit is specifically used for: Extracting the feature vector of the longitude and latitude information through the decision tree model; Inputting the feature vector into each tree in the decision tree model to obtain a leaf node corresponding to each tree, wherein each leaf node corresponds to a constraint condition; Detecting whether the leaf node meets the corresponding constraint condition; Based on whether the leaf nodes meet the corresponding constraints, output the score value corresponding to each tree.

10. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the data processing method according to any one of claims 1 to 6 are implemented.

11. A storage medium, characterized in that: A computer program is stored thereon, wherein when the computer program is executed by a processor, the steps of the data processing method according to any one of claims 1 to 6 are implemented.

Images