WO2023273298A1

WO2023273298A1 - User track recognition method, apparatus and device, and storage medium

Info

Publication number: WO2023273298A1
Application number: PCT/CN2022/071481
Authority: WO
Inventors: 张霖; 徐赛奕; 朱磊; 赵文婕
Original assignee: 平安科技（深圳）有限公司
Priority date: 2021-06-30
Filing date: 2022-01-12
Publication date: 2023-01-05
Also published as: CN113177101A; CN113177101B

Abstract

The present invention a user track recognition method, apparatus and device (800), and a storage medium, which relate to the field of data processing. The method comprises: obtaining original WiFi data and GPS information of a user in a time period to be recognized (101); performing data pre-processing on the original WiFi data to obtain data to be recognized (102); according to a preset expert rule dictionary, performing primary recognition on the data to be recognized to obtain a primary recognition result (103); if the primary recognition result is recognition failure, inputting the data to be recognized into a pretrained WiFi recognition model and obtaining a secondary recognition result (105); according to the primary recognition result or the secondary recognition result, generating user position annotation information of the user in the time period to be recognized; and generating a user track of the user according to the user position annotation information, the original WiFi data, and the GPS information (107). According to the present method, the user track of a user can be automatically recognized by means of a pre-established expert rule dictionary and a model.

Description

User trajectory recognition method, device, equipment and storage medium

This application claims the priority of the Chinese patent application with the application number 202110732370.4 and the title of the invention "User Trajectory Identification Method, Device, Equipment, and Storage Medium" submitted to the China Patent Office on June 30, 2021, the entire contents of which are incorporated by reference in application.

technical field

The present application relates to the field of data processing, and in particular to a user trajectory identification method, device, equipment and storage medium.

Background technique

The rapid development of smart terminals and positioning technology has greatly promoted the popularity of location-based service applications. Nowadays, users are the core basis for many companies to provide services. By analyzing the changes in user locations, user behavior can be described, which is useful for optimizing user recommendation systems. It is of great significance to improve the service quality of enterprises and help the layout of smart cities. Considering that the user's daily movement trajectory contains the user's information in time and space, which is closely related to the user's daily behavior, the research on user trajectory has always attracted the attention of scholars.

At present, the main methods used in user trajectory identification are mobile phone GPS identification and mobile phone base station identification. At present, there are the following deficiencies in identifying user trajectories through mobile phone GPS and base stations. First, due to the error of 0-100 meters in the existing GPS and base station due to the signal quality, the user trajectory judgment is wrong. Second, there will be multiple POIs (Point of Interest) at the same address or location, and it is impossible to accurately determine the actual trajectory of the user.

Contents of the invention

The present application provides a user trajectory identification method, device, equipment and storage medium, which are used to solve the technical problem that the existing user trajectory identification method has low accuracy in identifying the user's actual trajectory.

The first aspect of the present application provides a user trajectory identification method, including: obtaining the original wifi data and gps information of the user in the time period to be identified, the original wifi data including wifi connection time; performing data pre-processing on the original wifi data processing to obtain the data to be identified; according to the preset expert rule dictionary, the data to be identified is identified once to obtain a recognition result; if the identification result of the first identification is successful, the location category of the data to be identified is obtained ; if the primary recognition result is a recognition failure, then input the data to be recognized into a pre-trained wifi recognition model to obtain a secondary recognition result, wherein the secondary recognition result includes the data to be recognized location category; divide the section to be identified into slices according to the wifi connection time to obtain at least one section of wifi connection time section, and mark the wifi connection time section according to the location category of the data to be identified to obtain the user location labeling information; generating the user track of the user according to the user location labeling information, the original wifi data and the gps information. The second aspect of the present application provides a user trajectory identification method device, including a memory, a processor, and computer-readable instructions stored on the memory and operable on the processor, and the processor executes the computer The following steps are realized during the readable instruction: obtain the original wifi data and gps information of the user in the time period to be identified, the original wifi data includes the wifi connection time; the original wifi data is carried out data preprocessing to obtain the data to be identified; The preset expert rule dictionary is used to identify the data to be identified once to obtain an identification result; if the identification result of the identification is successful, the location category of the data to be identified is obtained; if the identification result of the identification is If the recognition fails, the data to be recognized is input into a pre-trained wifi recognition model to obtain a secondary recognition result, wherein the secondary recognition result includes the location category of the data to be recognized; according to the wifi connection Slicing and dividing the section to be identified by time to obtain at least one wifi connection time section, and marking the wifi connection time section according to the location category of the data to be identified to obtain user location labeling information; according to the user location labeling information, the original wifi data and the gps information to generate the user track of the user. The third aspect of the present application provides a computer-readable storage medium, wherein computer instructions are stored in the computer-readable storage medium, and when the computer instructions are run on the computer, the computer is made to perform the following steps: obtain the The original wifi data and gps information of time period, described original wifi data comprises wifi connection time; Carry out data preprocessing to described original wifi data, obtain the data to be identified; According to the preset expert rule dictionary, to the described data to be identified Perform a recognition to obtain a recognition result; if the recognition is successful, the location category of the data to be recognized is obtained; if the recognition is failed, the data to be recognized is input to the In the trained wifi recognition model, a secondary recognition result is obtained, wherein the secondary recognition result includes the location category of the data to be recognized; according to the wifi connection time, the segment to be recognized is divided into slices to obtain at least A section of wifi connection time period, and mark the wifi connection time period according to the location category of the data to be identified to obtain user location label information; according to the user location label information, the original wifi data and the gps information , to generate the user track of the user. The fourth aspect of the present application provides a user trajectory identification method and device, wherein the user trajectory identification method and device includes: an acquisition module that acquires the original wifi data and gps information of the user in the time period to be identified, the original wifi data includes wifi connection time; a preprocessing module for performing data preprocessing on the original wifi data to obtain data to be identified; a primary identification module for performing a primary identification on the data to be identified according to a preset expert rule dictionary, Obtain a primary recognition result, when the primary recognition result is a successful recognition, then obtain the location category of the data to be recognized; a secondary recognition module is used to convert the to-be-recognized data when the primary recognition result is a recognition failure The data is input into a pre-trained wifi recognition model to obtain a secondary recognition result, wherein the secondary recognition result includes the location category of the data to be recognized; a labeling module is used to classify the described wifi connection time according to the wifi connection time Sections to be identified are divided into slices to obtain at least one wifi connection time period, and the wifi connection time period is marked according to the location category of the data to be identified to obtain user position labeling information; the track drawing module is used for according to the described The user location label information, the original wifi data and the gps information generate the user track of the user. In the technical solution provided by the present application, the original wifi data and gps information of the user in the time period to be identified are obtained, the original wifi data includes the wifi connection time; data preprocessing is performed on the original wifi data to obtain the data to be identified; according to the preset Expert rule dictionary, conduct a recognition on the data to be recognized, and get a recognition result; if the recognition result is successful, then get the location category of the data to be recognized; if the recognition result is a failure, then input the data to be recognized into the pre-training In a good wifi recognition model, the secondary recognition result is obtained, wherein the secondary recognition result includes the location category of the data to be recognized; according to the wifi connection time, the segment to be recognized is divided into slices to obtain at least one wifi connection time period, and according to the to-be-recognized data Identify the location category of the data and mark the wifi connection time period to obtain the user location label information; according to the user location label information, original wifi data and gps information, the user track of the user is generated. Based on deep learning technology, this method generates user location labeling information on wifi data, conducts small-scale fine user trajectory identification based on user location labeling information and wifi data, and combines GPS information to perform wide-area user trajectory identification in a large range to generate The user track of the user, the user track identification is performed according to various data, the recognition accuracy of the user track is improved, and the identification process of the user track can be automated.

Description of drawings

Fig. 1 is the schematic diagram of the first embodiment of the user trajectory identification method in the embodiment of the present application;

FIG. 2 is a schematic diagram of a second embodiment of the user trajectory identification method in the embodiment of the present application;

FIG. 3 is a schematic diagram of a third embodiment of the user trajectory identification method in the embodiment of the present application;

FIG. 4 is a schematic diagram of a fourth embodiment of the user trajectory identification method in the embodiment of the present application;

FIG. 5 is a schematic diagram of a fifth embodiment of the user trajectory identification method in the embodiment of the present application;

FIG. 6 is a schematic diagram of an embodiment of the user trajectory identification device in the embodiment of the present application;

FIG. 7 is a schematic diagram of another embodiment of the user trajectory identification device in the embodiment of the present application;

Fig. 8 is a schematic diagram of an embodiment of a user trajectory identification device in the embodiment of the present application.

detailed description

Embodiments of the present application provide a method, device, device, and storage medium for identifying user trajectories, which are used to solve the technical problem of low accuracy in identifying actual user trajectories in existing user trajectory identification methods.

The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and not necessarily Used to describe a specific sequence or sequence. It should be understood that the terms so used are interchangeable under appropriate circumstances such that the embodiments described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the term "comprising" or "having" and any variations thereof, are intended to cover a non-exclusive inclusion, for example, a process, method, system, product or device comprising a sequence of steps or elements is not necessarily limited to those explicitly listed instead, may include other steps or elements not explicitly listed or inherent to the process, method, product or apparatus.

For ease of understanding, the following describes the specific process of the embodiment of the present application. Please refer to FIG. 1. An embodiment of the user trajectory identification method in the embodiment of the present application includes:

101. Obtain the original wifi data and GPS information of the user in the time period to be identified;

It can be understood that the execution subject of the present application may be a user trajectory recognition device, and may also be a terminal or a server, which is not specifically limited here. The embodiment of the present application is described by taking the server as an execution subject as an example.

It should be emphasized that, in order to ensure data privacy and security, the above-mentioned original wifi data can be stored in a block chain node.

In this embodiment, the user track of the user in a time period can be identified, and the time period to be identified can be selected as one day or one week, which is not limited in this application.

In this embodiment, the original wifi data mainly includes wifi names, wifiids and connected wifi device numbers and corresponding wifi addresses of all wifis connected by the user in the time period to be identified, wherein each wifi name, wifiid and The device numbers are one-to-one, and after the subsequent category recognition of the wifi name, the identified category, wifi name, wifiid, and device number together form a user track.

In this embodiment, the outdoor user trajectory of the time period to be identified can be described through the GPS information, and indoor positioning can be performed through the original wifi data, and the indoor user trajectory of the user in the time period to be identified can be described, combined with the indoor User trajectories and outdoor user trajectories can obtain the total user trajectories of users.

102. Perform data preprocessing on the original wifi data to obtain data to be identified;

In practical applications, the original wifi data collected is sometimes not complete, sometimes not recorded, sometimes just write a number, there is also a feature that has data and other data that are all 0, these data do not meet the requirements of the algorithm itself , for example, in the regression model, the correlation and collinear features will cause the algorithm to fail to converge or fail, and need to be processed in advance; in this embodiment, data preprocessing mainly includes data cleaning and word segmentation, and data cleaning mainly deletes abnormal data , delete invalid data, delete blank data, word segmentation processing is mainly to segment the wifi name data in the original wifi data, decompose the wifi name into an array composed of multiple words, and remove stop words, stop words are composed of no actual Functional words of meaning, such as modal particles, punctuation, etc., and then delete these stop words from the wifi name array, and the rest are valid words, and the valid phrases are jointly constructed by valid words. The purpose of this is to reduce subsequent operations quantity.

In this embodiment, the wifi name is segmented mainly through the stuttering word segmentation method, which is a stuttering word segmentation module of Python, which supports three word segmentation modes: precise mode, full mode and search engine mode. The stop words in the wifi name data can be removed through the preset stop dictionary, and the number of stop words in the stop word dictionary can be increased according to different needs.

103. Perform one recognition on the data to be recognized according to the preset expert rule dictionary, and obtain one recognition result;

In this embodiment, for some specific words in the wifi name, it is to identify the category of the wifi name, such as including "airport" and "railway station", generally for transportation; including "restaurant" for food and beverage; including common router brand names Usually home wifi. For some special phrases, an expert rule dictionary is established, and the dictionary contains specific words corresponding to types and categories. In addition, for specific application scenarios, we can also create special dictionaries. For example, a car brand name dictionary is used to identify Wi-Fi in places such as car sales and car services; a dictionary of KTV, clubs, equestrian and other store names is used to identify Wi-Fi in high-end consumer places. Traversing the wifi name array output in the previous step, if it hits a rule word in the dictionary, the wifi name will be marked as the specified category. Expert identification rules can classify and identify most wifi names containing keywords, effectively reducing the amount of subsequent modeling data, and do special rule screening for wifi in specific places to improve the accuracy of identification.

104. If the recognition result is successful, the location category of the data to be recognized is obtained;

105. If the primary recognition result is a recognition failure, input the data to be recognized into the pre-trained wifi recognition model to obtain a secondary recognition result, the secondary recognition result includes the location category of the data to be recognized;

In this embodiment, a recognition is performed through the expert rule dictionary, and the array of wifi names output in the previous step is traversed through the expert rule dictionary. If none of the rule words in the dictionary are hit, it is determined that the first recognition fails, and a second recognition is required. Secondary recognition is carried out through the preset wifi recognition model, wherein the wifi recognition model is established through automatic training by using the ability of machine automatic learning and through the already marked training data.

In this embodiment, the wifi recognition model is formed using DNN (Deep Neural Network), and the DNN model includes a word vector layer, a maximum pooling layer, a fully connected hidden layer and an output layer, and the words in the wifi name array are mapped by the word vector layer To the word vector, the maximum pooling is performed on the time series. The pooling process eliminates the difference in the number of words in different corpus samples, and extracts the maximum value of each subscript position in the word vector. After the maximum pooling The vector is sent to two consecutive fully connected hidden layers for calculation, and finally a normalized probability distribution is output through the output layer, and the result is 1. The output layer has multiple neurons, and the number of neurons is the same as the wifi needs to be classified. The categories are the same, so the output of the xth neuron can be considered as the predicted probability that the wifi name data belongs to the xth category of wifi category, and the wifi category corresponding to the maximum value is used as the category of the wifi name data, and as the secondary recognition result.

106. Slicing the section to be identified according to the wifi connection time to obtain at least one section of the wifi connection time period, and marking the wifi connection time period according to the location category of the data to be identified to obtain user location labeling information;

In this embodiment, the primary recognition result is obtained through the expert rule dictionary, and the secondary recognition result is obtained through the wifi recognition model. Both the primary recognition result and the secondary recognition result have the recognition type of the original wifi data, and the recognized type The wifi name data is associated with the wifi name row that the user is connected to, so that the type of wifi that the user is connected to can be identified. By summarizing the types of wifi connected by the user, the user location label information of the user can be identified. The identified user location label information is as follows:

0-8 o'clock home wifiid: aaabac device number xxxxx;

9 o'clock traffic travel wifiid: erdfhethrh device number xYYYY;

From 10:00 to 12:00, wifiid of office buildings and commercial buildings: qegehwr equipment number xYYX;

Restaurant wifiid at 13:00: EHFDrh device number xxxyzz;

14-18 o'clock office building commercial building wifiid: ETHHDF equipment number xxxyzzz;

18:00-22:00 entertainment facility wifiid: ehdfhR device number xxxyzzzz;

From 22 o'clock to 24 o'clock family home wifiid: erhDSG device number xxxyyzzzz.

107. Generate a user track of the user according to the user location label information, original wifi data, and GPS information.

In this embodiment, the user's GPS information can be used to locate the user, and then realize the generation of the user's track. However, due to the serious attenuation of the signal and the multipath effect, the GPS cannot work effectively in the building, and there are 0-100 The error of the meter causes the wrong judgment of the user's trajectory. When indoors, the wifi address information in the original wifi data can be used to describe the user's indoor user trajectory, mainly through the method of location fingerprinting. The location of each location is associated with some kind of "fingerprint", and a location corresponds to a unique fingerprint. This fingerprint can be single-dimensional or multi-dimensional. For example, if the device to be located is receiving or sending information, then the fingerprint can be one or more characteristics of this information or signal, such as the signal strength and delay of the connected wifi, etc., through wifi Address and location fingerprints are used for indoor positioning of users, combined with GPS outdoor positioning for all-round precise positioning.

In this embodiment, after combining the user's indoor and outdoor trajectories, the locations connected to wifi in the trajectories are marked in conjunction with the user's location labeling information. Through the marked information and trajectories, it can be judged whether the user deviates from the daily trajectories. If the user deviates from the daily trajectories, Then remind the person associated with the user in advance. For example, if the user is a student and the user's trajectory deviates from the daily trajectory during the time period to be identified, the parents can be reminded. In this embodiment, the type of wifi that the user is connected to can be identified through the user location annotation information of the user track, and if it is not the type of wifi that is connected daily, it can be determined that the user deviates from the daily track.

In this embodiment, the original wifi data and gps information of the user in the time period to be identified are obtained, and the original wifi data includes the wifi connection time; data preprocessing is carried out to the original wifi data to obtain the data to be identified; according to the preset expert rule dictionary , perform a recognition on the data to be recognized, and get a recognition result; if the recognition result is successful, then get the location category of the data to be recognized; if the recognition result is a failure, then input the data to be recognized to the pre-trained wifi In the recognition model, a secondary recognition result is obtained, wherein the secondary recognition result includes the location category of the data to be recognized; according to the wifi connection time, the segment to be recognized is divided into slices to obtain at least one wifi connection time period, and according to the data to be recognized The location category marks the wifi connection time period to obtain the user location label information; according to the user location label information, original wifi data and gps information, the user's user track is generated. Based on deep learning technology, this method generates user location labeling information on wifi data, conducts small-scale fine user trajectory identification based on user location labeling information and wifi data, and combines GPS information to perform wide-area user trajectory identification in a large range to generate The user track of the user, the user track identification is performed according to various data, the recognition accuracy of the user track is improved, and the identification process of the user track can be automated.

Please refer to Fig. 2, the second embodiment of the user trajectory identification method in the embodiment of the present application includes:

201. Obtain the original wifi data and GPS information of the user in the time period to be identified;

Step 201 in this embodiment is similar to step 101 in the first embodiment, and will not be repeated here.

202. Perform data cleaning processing on the wifi name data to obtain a data cleaning result;

In this embodiment, in this embodiment, data preprocessing mainly includes data cleaning and word segmentation, wherein data cleaning mainly deletes abnormal data, deletes invalid data, and deletes blank data. Word segmentation of the wifi name data, decomposing the wifi name into an array composed of multiple words, and removing stop words, which are composed of functional words without actual meaning, such as modal particles, punctuation, etc., and then from the wifi name array These stop words are deleted, and the rest are effective words, and the effective phrases are jointly constructed by the effective words. The purpose of this is to reduce the amount of subsequent calculations.

203. Segment the wifi name data in the data cleaning result into single characters to obtain a sequence array;

204. According to the preset prefix dictionary, construct the directed acyclic graph of the sequence array, and calculate the probability of each path in the directed acyclic graph respectively;

205. Obtain the optimal word segmentation result according to the path corresponding to the maximum probability in the directed acyclic graph, and perform word segmentation on the wifi name data in the data cleaning result according to the optimal word segmentation result to obtain a wifi word segmentation array;

206. Eliminate the stop words in the wifi word segmentation array to obtain the data to be identified;

In this embodiment, the prefix dictionary is constructed based on the statistical dictionary. For example, the prefixes of the word "Peking University" in the statistical dictionary are "北", "北京", and "北京大" respectively; the prefixes of the word "大学" are "大" , use "北", "北京", "北京大" and "大" as prefixes, and then segment the input text based on the prefix dictionary. For "go", there is no prefix, so there is only one division method; for " For "North", there are three ways of dividing "North", "Beijing" and "Peking University"; for "Beijing", there is only one way of dividing; for "Da", there are two ways of dividing "Da" and "University" The division method, and so on, can obtain the division method of the prefix word at the beginning of each character. If the string to be segmented has m characters, considering the left and right positions of each character, there are m+1 points corresponding, and the number of points is from 0 to m. Considering candidate words as edges, a word segmentation graph can be generated based on the dictionary.

In jieba participle, the frequency of each word will be marked (equal to the number of occurrences divided by the total number, when the overall sample is large, it can be approximated as the probability of the word), after knowing the frequency of each word, it can be based on The dynamic programming method is used to find the word segmentation path with the highest probability. General dynamic programming finds the optimal path from left to right, but here it is from right to left to find the optimal path. This is mainly because the center of gravity in a Chinese sentence is often at the back, which is the backbone of the sentence, so the correct rate calculated from right to left is often higher than that from left to right.

207. Perform one recognition on the data to be recognized according to the preset expert rule dictionary, and obtain one recognition result;

208. If the result of one recognition is a successful recognition, the location category of the data to be recognized is obtained;

209. If the primary recognition result is a recognition failure, input the data to be recognized into the pre-trained wifi recognition model to obtain a secondary recognition result, the secondary recognition result includes the location category of the data to be recognized;

210. Divide the section to be identified into slices according to the wifi connection time, obtain at least one section of the wifi connection time period, and mark the wifi connection time period according to the location category of the data to be identified, to obtain user location label information;

211. Generate a user track of the user according to the user location label information, original wifi data, and GPS information.

Steps 207-211 in this embodiment are similar to steps 103-107 in the first embodiment, and will not be repeated here.

On the basis of the previous embodiment, this embodiment describes in detail the process of performing data preprocessing on the original wifi data to obtain the data to be identified, and obtains the data cleaning result by performing data cleaning on the wifi name data; The wifi name data in the data cleaning result is subjected to word segmentation processing to obtain a wifi word segmentation array; the stop words in the wifi word segmentation array are removed to obtain the data to be identified. Through data cleaning in data preprocessing, word segmentation and stop word elimination, the remaining effective words are used to construct effective phrases together, reducing the amount of follow-up calculations and improving the efficiency of user trajectory recognition.

Please refer to Fig. 3, the third embodiment of the user trajectory identification method in the embodiment of the present application includes:

301. Obtain the original wifi data and GPS information of the user in the time period to be identified;

302. Perform data preprocessing on the original wifi data to obtain data to be identified;

Steps 301-302 in this embodiment are similar to steps 101-102 in the first embodiment, and will not be repeated here.

303. Match the data to be identified with the location words in the expert rule dictionary;

304. If the matching is successful, the location category corresponding to the location word whose data to be identified is successfully matched is used as a recognition result;

305. If the matching fails, set a recognition result as recognition failure;

In this embodiment, in this embodiment, for some specific words in the wifi name, it is to identify the category of the wifi name, such as including "airport" and "railway station", which are generally transportation trips; including "restaurant" are generally catering; Including the common router brand name is generally home wifi. For some special phrases, an expert rule dictionary is established, and the dictionary contains specific words corresponding to types and categories. In addition, for specific application scenarios, we can also create special dictionaries. For example, a car brand name dictionary is used to identify Wi-Fi in places such as car sales and car services; a dictionary of KTV, clubs, equestrian and other store names is used to identify Wi-Fi in high-end consumer places. Expert identification rules can classify and identify most wifi names containing keywords, effectively reducing the amount of subsequent modeling data, and do special rule screening for wifi in specific places to improve the accuracy of identification. By traversing the wifi name array output in the previous step, if the regular word in the dictionary is hit, the wifi name will be marked as the specified category. If none of the regular words in the dictionary are hit, it is determined that the first recognition fails, and a second recognition is required. .

306. If the primary recognition result is a recognition failure, input the data to be recognized into the pre-trained wifi recognition model to obtain a secondary recognition result, the secondary recognition result includes the location category of the data to be recognized;

307. Divide the section to be identified into slices according to the wifi connection time, obtain at least one section of the wifi connection time period, and mark the wifi connection time period according to the location category of the data to be identified, to obtain user location label information;

308. Generate a user track of the user according to the user location label information, original wifi data, and GPS information.

Steps 306-308 in this embodiment are similar to steps 104-106 in the first embodiment, and will not be repeated here.

On the basis of the previous embodiments, this embodiment describes in detail that according to the preset expert rule dictionary, the data to be recognized is recognized once to obtain a recognition result. By combining the data to be recognized with the expert rule dictionary If the matching is successful, the location category corresponding to the location word that the data to be recognized is successfully matched is used as a recognition result; if the matching fails, the recognition result is set as a recognition failure. Through the preset expert rule dictionary, the data to be recognized is roughly recognized, and the data to be recognized that is successfully recognized does not need to be re-recognized, which improves the efficiency of user trajectory recognition.

Please refer to Fig. 4, the fourth embodiment of the user trajectory identification method in the embodiment of the present application includes:

401. Obtain historical wifi data and a preset neural network model, and initialize the network parameters of the word vector layer, the maximum pooling layer and the network parameters of the fully connected hidden layer in the neural network model. Historical wifi data includes manually identified location categories ;

402. Input the historical wifi data into the neural network model to obtain the predicted location category;

403. Calculate a preset loss function according to the location category manually identified by the historical wifi data and the location category predicted by the neural network model, to obtain a loss value, and determine whether the loss value is less than a preset threshold;

404. If so, then determine the wifi recognition model according to the network parameters of the word vector layer, the maximum pooling layer and the fully connected hidden layer in the neural network model;

405. If not, update the network parameters of the neural network model through the backpropagation algorithm according to the loss value, iterate the model training process repeatedly until the loss value is less than the preset threshold, and determine the middle word vector layer, The network parameters of the maximum pooling layer sum and the network parameters of the fully connected hidden layer determine the wifi recognition model;

In this embodiment, the word vector model (word to vector, word2vec) is used to pre-train historical wifi data to obtain the trained word vector weight, and the word vector layer and convolution layer of the neural network model are initialized using the word vector weight. Parameters of network layer and fully connected layer. It is set not to update the parameters of the word vector layer before the nth training round, n is a positive integer, and the specific value is set by the technician according to the actual situation.

In this embodiment, based on the parameters of the loss value convolutional network layer and the fully connected hidden layer, the parameters of the word vector layer are adjusted from the n+1 round of training. Adjusting the learning rate of other network layers of the neural network model except the word vector layer to the initial learning rate, adjusting the learning rate of the word vector layer to a learning rate smaller than the initial learning rate, continuing to train the neural network model until convergence, That is, if the loss value is less than the error threshold, it is determined that the converged neural network model is the wifi recognition model.

406. Obtain the original wifi data and GPS information of the user in the time period to be identified;

407. Perform data preprocessing on the original wifi data to obtain data to be identified;

408. Perform one recognition on the data to be recognized according to the preset expert rule dictionary, and obtain one recognition result;

409. If the result of a recognition is a successful recognition, the location category of the data to be recognized is obtained;

Steps 406-409 in this embodiment are similar to steps 101-104 in the first embodiment, and will not be repeated here.

410. If a recognition result is a recognition failure, input the data to be recognized into the word vector layer in the wifi recognition model, and convert the data to be recognized into a word vector sequence;

411. Input the word vector into the maximum pooling layer in the wifi recognition model to obtain the maximum pooling result;

412. Input the maximum pooling result into the fully connected hidden layer in the wifi recognition model, and use the Softmax function to classify the output results of the fully connected hidden layer to obtain the secondary recognition result, which includes the data to be recognized location category;

In this embodiment, in this embodiment, the wifi recognition model includes a word vector layer, a maximum pooling layer, a fully connected hidden layer and an output layer, wherein the word vector layer is used to better represent the semantics between different words relationship, first transform words into fixed-dimensional vectors. After the training is completed, the semantic similarity between words and words can be expressed by the distance between their word vectors. The more semantically similar, the closer the distance. The maximum pooling layer is used to eliminate the difference in the number of words in different corpus samples. difference, and extract the maximum value at each subscript position in the word vector. After pooling, the vector sequence output by the word vector layer is converted into a fixed-dimensional vector. For example, assuming that the sequence of vectors before max pooling is [[2, 3, 5], [7, 3, 6], [1, 4, 0]], the result of max pooling is: [7, 4, 6]. The fully connected hidden layer is used to send the vector after the maximum pooling to two consecutive hidden layers for calculation. There is a fully connected structure between the hidden layers, and the number of neurons in the output layer is consistent with the number of categories of samples. For example, in In the binary classification problem, the output layer will have 2 neurons. Through the Softmax activation function, the output result is a normalized probability distribution, and the sum is 1. This application is a multi-classification model, and the output layer has multiple neurons. The number of neurons is the same as the category that wifi needs to classify, so the xth The output of each neuron can be considered as the predicted probability that the wifi name data belongs to the x-th wifi category, and the wifi category corresponding to the maximum value is used as the category of the wifi name data, and as the secondary recognition result.

413. Divide the section to be identified into slices according to the wifi connection time, obtain at least one section of the wifi connection time period, and mark the wifi connection time period according to the location category of the data to be identified, to obtain user location label information;

414. Generate a user track of the user according to the user location label information, original wifi data, and GPS information.

Based on the previous embodiments, this embodiment describes in detail the process of inputting data to be recognized into a pre-trained wifi recognition model to obtain a secondary recognition result. By inputting the data to be recognized into the word vector layer in the wifi recognition model, the data to be recognized is converted into a sequence of word vectors; the word vector is input into the maximum pooling layer in the wifi recognition model to obtain the maximum pooling result; the maximum pooling Input the results of the transformation into the fully connected hidden layer in the wifi recognition model, and use the Softmax function to classify the output results of the fully connected hidden layer to obtain the secondary recognition results, wherein the secondary recognition results include the location corresponding to the data to be recognized category. At the same time, the specific training process of the wifi recognition model is described in detail. The wifi recognition model constructed by deep learning can accurately determine the location category of wifi data, thereby improving the generation efficiency of user trajectories.

Please refer to Fig. 5, the fifth embodiment of the user trajectory identification method in the embodiment of the present application includes:

501. Obtain the original wifi data and GPS information of the user in the time period to be identified;

502. Perform data preprocessing on the original wifi data to obtain data to be identified;

503. Perform one recognition on the data to be recognized according to the preset expert rule dictionary, and obtain one recognition result;

504. If the result of one recognition is a successful recognition, the location category of the data to be recognized is obtained;

505. If the primary recognition result is a recognition failure, input the data to be recognized into a pre-trained wifi recognition model to obtain a secondary recognition result;

Steps 501-505 in this embodiment are similar to steps 101-105 in the first embodiment, and will not be repeated here.

506. According to the first recognition result or the second recognition result, obtain the location category of all the original wifi data in the time period to be recognized;

507. Slicing and dividing the time period to be identified according to the wifi connection time of the original wifi data, to obtain at least one wifi connection time period;

508. According to the wifi connection time period corresponding to all the original wifi data and the location category corresponding to all the original wifi data, generate user location labeling information for the time period to be identified;

In this embodiment, the user may have connected to multiple wifis during the time period to be identified, and there may be an intermediate interval between multiple wifi connections. For example, the time period to be identified is when the user connects to wifi on a certain day. The time is 0-8:00-9:00-10:00-12:00-13:00-14-18:00-18:00-22:00-22:00-24:00. The connection time is stored as one of the original wifi data, and the identified type of wifi name data is associated with the wifi name row connected by the user, so that the type of wifi connected by the user can be identified. By summarizing the types of wifi connected by the user, the user location label information of the user can be identified. The identified user location label information is as follows:

0-8 o'clock home wifiid: aaabac device number xxxxx;

9 o'clock traffic travel wifiid: erdfhethrh device number xYYYY;

Restaurant wifiid at 13:00: EHFDrh device number xxxyzz;

18:00-22:00 entertainment facility wifiid: ehdfhR device number xxxyzzzz;

509. Generate a user track of the user according to the user location label information, original wifi data, and GPS information.

Step 509 in this embodiment is similar to step 107 in the first embodiment, and will not be repeated here.

On the basis of the previous embodiments, this embodiment describes in detail how to generate the user location label information of the user in the time period to be recognized according to the primary recognition result or the secondary recognition result. Obtain the location categories of all original wifi data in the time period to be identified through the primary recognition result or the secondary recognition result; segment the time period to be recognized according to the wifi connection time of the original wifi data, and obtain at least one wifi connection time period; according to The wifi connection time periods corresponding to all the original wifi data and the location categories corresponding to all the original wifi data are used to generate user location label information for the time period to be identified. Through this method, the location type of the wifi data can be marked, and then the location type of each point in the user track can be generated.

The user trajectory identification method in the embodiment of the present application is described above, and the user trajectory identification device in the embodiment of the application is described below. Please refer to FIG. 6. An embodiment of the user trajectory identification device in the embodiment of the application includes:

Obtaining module 601, obtaining the original wifi data and gps information of the user in the time period to be identified, the original wifi data including wifi connection time;

A preprocessing module 602, configured to perform data preprocessing on the original wifi data to obtain data to be identified;

The one-time recognition module 603 is configured to perform one-time recognition on the data to be recognized according to the preset expert rule dictionary, and obtain a recognition result once, and when the recognition result of the first recognition is successful, obtain the location of the data to be recognized category;

The secondary recognition module 604 is configured to input the data to be recognized into a pre-trained wifi recognition model to obtain a secondary recognition result when the primary recognition result is a recognition failure, wherein the secondary recognition result a category of locations comprising said data to be identified;

A labeling module 605, configured to divide the section to be identified into slices according to the wifi connection time to obtain at least one section of the wifi connection time period, and mark the wifi connection time period according to the location category of the data to be identified, Obtain user location annotation information;

A trajectory drawing module 606, configured to generate a user trajectory of the user according to the user location annotation information, the original wifi data and the GPS information.

In the embodiment of the present application, the user trajectory identification device operates the above user trajectory identification method, the user trajectory identification device obtains the original wifi data and gps information of the user in the time period to be identified, the original wifi data includes wifi connection time; for the original Perform data preprocessing on the wifi data to obtain the data to be recognized; according to the preset expert rule dictionary, perform a recognition on the data to be recognized once, and obtain a recognition result; if the recognition result is successful, then obtain the location category of the data to be recognized; If the primary recognition result is recognition failure, input the data to be recognized into the pre-trained wifi recognition model to obtain the secondary recognition result, wherein the secondary recognition result includes the location category of the data to be recognized; Sections are divided into slices to obtain at least one wifi connection time period, and the wifi connection time period is marked according to the location category of the data to be identified to obtain the user location label information; according to the user location label information, original wifi data and gps information, the user is generated user trajectory. Based on deep learning technology, this method generates user location labeling information for wifi data, conducts small-scale fine user trajectory identification based on user location labeling information and wifi data, and combines GPS information for wide-area user trajectory identification to generate The user track of the user, the user track identification is performed according to various data, the recognition accuracy of the user track is improved, and the identification process of the user track can be automated.

Please refer to FIG. 7, the second embodiment of the user trajectory recognition device in the embodiment of the present application includes:

Wherein, the preprocessing module 602 includes: a data cleaning unit 6021, which is used to perform data cleaning processing on the wifi name data to obtain a data cleaning result; a word segmentation unit 6022, which is used to convert the wifi name data in the data cleaning result Perform word segmentation processing to obtain a wifi word segmentation array; the removing unit 6023 is used to remove stop words in the wifi word segmentation array to obtain data to be recognized.

Optionally, the word segmentation unit 6022 is specifically configured to: perform word segmentation on the wifi name data in the data cleaning result to obtain a sequence array; construct a directed non-return sequence array of the sequence array according to a preset prefix dictionary. , and calculate the probability of each path in the directed acyclic graph respectively; according to the path corresponding to the maximum probability in the directed acyclic graph, the optimal word segmentation result is obtained, and the optimal word segmentation result is used for the described optimal word segmentation result The wifi name data in the data cleaning result is segmented to obtain a wifi word segmentation array.

Optionally, the primary recognition module 603 is specifically configured to: match the data to be recognized with the location words in the expert rule dictionary; The category of the location is used as a recognition result; if the matching fails, the recognition result is set as a recognition failure.

Optionally, the secondary recognition module 604 is specifically configured to: input the data to be recognized into the word vector layer in the wifi recognition model, convert the data to be recognized into a word vector sequence; convert the word The vector is input to the maximum pooling layer in the wifi recognition model to obtain the maximum pooling result; the maximum pooling result is input to the fully connected hidden layer in the wifi recognition model, and by the Softmax function, the The output results of the fully connected hidden layer are classified to obtain the secondary recognition result.

Optionally, the user trajectory identification device further includes a model training module 607, and the model training module 607 is specifically configured to: acquire historical wifi data and a preset neural network model, and initialize the word vector layer in the neural network model , the network parameters of the maximum pooling layer and the network parameters of the fully connected hidden layer, the historical wifi data includes the artificially identified place category; the historical wifi data is input in the neural network model to obtain the predicted place category; Calculate the preset loss function according to the location category manually identified and the location category predicted by the neural network model according to the historical wifi data, obtain a loss value, and judge whether the loss value is less than a preset threshold; if so, then according to the specified The network parameters of the word vector layer, the maximum pooling layer and the fully connected hidden layer in the neural network model determine the wifi identification model; if not, then update the network parameters of the neural network model by the backpropagation algorithm according to the loss value, Iterate the model training process repeatedly until the loss value is less than the preset threshold, and determine the network parameters of the word vector layer, the maximum pooling layer and the network parameters of the fully connected hidden layer of the trained neural network model to determine the wifi recognition model.

Optionally, the labeling module 606 is specifically configured to: slice and divide the time period to be identified according to the wifi connection time of the original wifi data to obtain at least one period of wifi connection time; The corresponding relationship between each original wifi data and each data to be identified determines the location category corresponding to each original wifi data; according to the location category of the original wifi data and the original wifi data, the wifi connection time period is marked to obtain the user Location annotation information.

On the basis of the previous embodiment, this embodiment describes in detail the specific functions of each module and the unit composition of some modules. Through the newly added module, based on deep learning technology, the user location labeling information is generated for wifi data. According to The user location tagging information and wifi data are used for small-scale fine user trajectory identification, combined with GPS information for wide-area wide-area user trajectory identification, to generate the user trajectory of the user, and to perform user trajectory identification based on various data to improve The recognition accuracy of user trajectories can automate the identification process of user trajectories.

The above Figures 6 and 7 describe in detail the user trajectory identification device in the embodiment of the present application from the perspective of modular functional entities, and the following describes the user trajectory identification device in the embodiment of the application in detail from the perspective of hardware processing.

Fig. 8 is a schematic structural diagram of a user trajectory recognition device provided by an embodiment of the present application. The user trajectory recognition device 800 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units) , CPU) 810 (eg, one or more processors) and memory 820, and one or more storage media 830 (eg, one or more mass storage devices) for storing application programs 833 or data 832 . Wherein, the memory 820 and the storage medium 830 may be temporary storage or persistent storage. The program stored in the storage medium 830 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations for the user trajectory recognition device 800 . Furthermore, the processor 810 may be configured to communicate with the storage medium 830, and execute a series of instruction operations in the storage medium 830 on the user trajectory identification device 800, so as to implement the steps of the above user trajectory identification method.

The user trajectory recognition device 800 may also include one or more power sources 840, one or more wired or wireless network interfaces 850, one or more input and output interfaces 860, and/or, one or more operating systems 831, such as Windows Server , Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art can understand that the structure of the user trajectory recognition device shown in FIG. 8 does not constitute a limitation to the user trajectory recognition device provided in this application, and may include more or less components than those shown in the figure, or combine certain components, or different component arrangements.

The blockchain referred to in this application is a new application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain (Blockchain), essentially a decentralized database, is a series of data blocks associated with each other using cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify its Validity of information (anti-counterfeiting) and generation of the next block. The blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

The present application also provides a computer-readable storage medium. The computer-readable storage medium may be a non-volatile computer-readable storage medium. The computer-readable storage medium may also be a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium, and when the instructions are run on the computer, the computer is made to execute the steps of the user trajectory identification method.

Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working process of the system, device, and unit described above can refer to the corresponding process in the foregoing method embodiments, and details are not repeated here.

If the integrated unit is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or part of the contribution to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , including several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (read-only memory, ROM), random access memory (random access memory, RAM), magnetic disk or optical disc and other media that can store program codes. .

As mentioned above, the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still understand the foregoing The technical solutions described in each embodiment are modified, or some of the technical features are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the application.

Claims

A user trajectory identification method, wherein the user trajectory identification method comprises:

Obtain the original wifi data and gps information of the user in the time period to be identified, wherein the original wifi data includes wifi connection time;

Carry out data preprocessing to described original wifi data, obtain the data to be identified;

performing a recognition on the data to be recognized according to a preset dictionary of expert rules to obtain a recognition result;

If the recognition result of the first recognition is successful, the location category of the data to be recognized is obtained;

If the primary recognition result is a recognition failure, then input the data to be recognized into a pre-trained wifi recognition model to obtain a secondary recognition result, wherein the secondary recognition result includes the location of the data to be recognized category;

Slicing and dividing the segment to be identified according to the wifi connection time to obtain at least one wifi connection time segment, and marking the wifi connection time segment according to the location category of the data to be identified to obtain user location tagging information;

Generating the user track of the user according to the user location tag information, the original wifi data and the gps information.
The user track identification method according to claim 1, wherein the original wifi data includes wifi name data, and performing data preprocessing on the original wifi data to obtain data to be identified includes:

Perform data cleaning processing on the wifi name data to obtain a data cleaning result;

Carry out word segmentation processing to wifi name data in the data cleaning result, obtain wifi word segmentation array;

The stop words in the wifi word segmentation array are removed to obtain the data to be recognized.
The user track identification method according to claim 2, wherein said wifi name data in said data cleaning result is subjected to word segmentation processing to obtain a wifi word segmentation array comprising:

Carry out word segmentation to the wifi name data in the data cleaning result, obtain sequence array;

Constructing the directed acyclic graph of the sequence array according to the preset prefix dictionary, and calculating the probability of each path in the directed acyclic graph respectively;

According to the path corresponding to the maximum probability in the directed acyclic graph, an optimal word segmentation result is obtained, and according to the optimal word segmentation result, the wifi name data in the data cleaning result is segmented to obtain a wifi word segmentation array.
The user trajectory recognition method according to any one of claims 1-3, wherein said performing one recognition on said to-be-recognized data according to a preset expert rule dictionary, and obtaining a recognition result comprises:

Matching the data to be identified with the location words in the expert rule dictionary;

If the matching is successful, the location category corresponding to the location word that the data to be identified is successfully matched is used as a recognition result;

If the matching fails, set the first recognition result as recognition failure.
The user trajectory identification method according to claim 4, wherein said inputting said data to be identified into a pre-trained wifi identification model to obtain a secondary identification result comprises:

The data to be identified is input to the word vector layer in the wifi recognition model, and the data to be identified is converted into a word vector sequence;

The word vector is input to the maximum pooling layer in the wifi recognition model to obtain the maximum pooling result;

The maximum pooling result is input into the fully connected hidden layer in the wifi identification model, and the output result of the fully connected hidden layer is classified by the Softmax function to obtain the secondary identification result.
The user track identification method according to claim 5, wherein the wifi identification model is obtained through the following steps of training:

Obtain historical wifi data and a preset neural network model, and initialize the network parameters of the word vector layer, the maximum pooling layer and the network parameters of the fully connected hidden layer in the neural network model, and the historical wifi data includes artificially identified location category;

The historical wifi data is input in the neural network model to obtain the predicted location category;

Calculate the preset loss function according to the location category manually identified and the location category predicted by the neural network model according to the historical wifi data, obtain a loss value, and judge whether the loss value is less than a preset threshold;

If so, then determine the wifi recognition model according to the network parameters of word vector layer, maximum pooling layer and fully connected hidden layer in the neural network model;

If not, then update the network parameters of the neural network model through the backpropagation algorithm according to the loss value, iteratively iterate the model training process until the loss value is less than the preset threshold, and determine the middle word vector of the trained neural network model Layer, the network parameters of the maximum pooling layer and the network parameters of the fully connected hidden layer determine the wifi recognition model.
The user trajectory identification method according to claim 5, wherein, according to the wifi connection time, the section to be identified is divided into slices to obtain at least one section of wifi connection time period, and according to the location category of the data to be identified The wifi connection time period is marked, and the user position marking information obtained includes:

Slicing and dividing the time period to be identified according to the wifi connection time of the original wifi data to obtain at least one wifi connection time period;

According to the corresponding relationship between each original wifi data and each data to be identified in the time period to be identified, determine the location category corresponding to each original wifi data;

Marking the wifi connection time period according to the location category of the original wifi data and the original wifi data to obtain user location marking information.
A user trajectory identification method device, comprising a memory, a processor, and computer-readable instructions stored on the memory and operable on the processor, and the processor implements the following steps when executing the computer-readable instructions :

Obtain the original wifi data and gps information of the user in the time period to be identified, wherein the original wifi data includes wifi connection time;

Carry out data preprocessing to described original wifi data, obtain the data to be identified;

performing a recognition on the data to be recognized according to a preset dictionary of expert rules to obtain a recognition result;

If the recognition result of the first recognition is successful, the location category of the data to be recognized is obtained;

If the primary recognition result is a recognition failure, then input the data to be recognized into a pre-trained wifi recognition model to obtain a secondary recognition result, wherein the secondary recognition result includes the location of the data to be recognized category;

Slicing and dividing the segment to be identified according to the wifi connection time to obtain at least one wifi connection time segment, and marking the wifi connection time segment according to the location category of the data to be identified to obtain user location tagging information;

Generating the user track of the user according to the user location tag information, the original wifi data and the gps information.
The user track identification method device according to claim 8, wherein the original wifi data includes wifi name data, and performing data preprocessing on the original wifi data to obtain data to be identified includes:

Perform data cleaning processing on the wifi name data to obtain a data cleaning result;

Carry out word segmentation processing to wifi name data in the data cleaning result, obtain wifi word segmentation array;

The stop words in the wifi word segmentation array are removed to obtain the data to be recognized.
The user track identification method device according to claim 9, wherein said performing word segmentation processing on the wifi name data in the data cleaning result, and obtaining the wifi word segmentation array includes:

Carry out word segmentation to the wifi name data in the data cleaning result, obtain sequence array;

Constructing the directed acyclic graph of the sequence array according to the preset prefix dictionary, and calculating the probability of each path in the directed acyclic graph respectively;

According to the path corresponding to the maximum probability in the directed acyclic graph, the optimal word segmentation result is obtained, and according to the optimal word segmentation result, the wifi name data in the data cleaning result is segmented to obtain the wifi word segmentation array.
According to the user trajectory identification method device according to any one of claims 8-10, wherein, performing a primary identification on the data to be identified according to a preset expert rule dictionary, and obtaining a primary identification result includes:

Matching the data to be identified with the location words in the expert rule dictionary;

If the matching is successful, the location category corresponding to the location word that the data to be identified is successfully matched is used as a recognition result;

If the matching fails, set the first recognition result as recognition failure.
The user trajectory identification method device according to claim 11, wherein said inputting said data to be identified into a pre-trained wifi identification model to obtain a secondary identification result comprises:

The data to be identified is input to the word vector layer in the wifi recognition model, and the data to be identified is converted into a word vector sequence;

The word vector is input to the maximum pooling layer in the wifi recognition model to obtain the maximum pooling result;

The maximum pooling result is input into the fully connected hidden layer in the wifi identification model, and the output result of the fully connected hidden layer is classified by the Softmax function to obtain the secondary identification result.
The user track identification method device according to claim 12, wherein the wifi identification model is obtained through the following steps of training:

Obtain historical wifi data and a preset neural network model, and initialize the network parameters of the word vector layer, the maximum pooling layer and the network parameters of the fully connected hidden layer in the neural network model, and the historical wifi data includes artificially identified location category;

The historical wifi data is input in the neural network model to obtain the predicted location category;

Calculate the preset loss function according to the location category manually identified and the location category predicted by the neural network model according to the historical wifi data, obtain a loss value, and judge whether the loss value is less than a preset threshold;

If so, then determine the wifi recognition model according to the network parameters of word vector layer, maximum pooling layer and fully connected hidden layer in the neural network model;

If not, then update the network parameters of the neural network model through the backpropagation algorithm according to the loss value, iteratively iterate the model training process until the loss value is less than the preset threshold, and determine the middle word vector of the trained neural network model Layer, the network parameters of the maximum pooling layer and the network parameters of the fully connected hidden layer determine the wifi recognition model.
The user trajectory identification method device according to claim 12, wherein, according to the wifi connection time, the section to be identified is divided into slices to obtain at least one section of wifi connection time period, and according to the location of the data to be identified The category marks the wifi connection time period, and the obtained user location marking information includes:

Slicing and dividing the time period to be identified according to the wifi connection time of the original wifi data to obtain at least one wifi connection time period;

According to the corresponding relationship between each original wifi data and each data to be identified in the time period to be identified, determine the location category corresponding to each original wifi data;

Marking the wifi connection time period according to the location category of the original wifi data and the original wifi data to obtain user location marking information.
A computer-readable storage medium, wherein computer instructions are stored in the computer-readable storage medium, and when the computer instructions are run on the computer, the computer is made to perform the following steps:

Obtain the original wifi data and gps information of the user in the time period to be identified, wherein the original wifi data includes wifi connection time;

Carry out data preprocessing to described original wifi data, obtain the data to be identified;

performing a recognition on the data to be recognized according to a preset dictionary of expert rules to obtain a recognition result;

If the recognition result of the first recognition is successful, the location category of the data to be recognized is obtained;

If the primary recognition result is a recognition failure, then input the data to be recognized into a pre-trained wifi recognition model to obtain a secondary recognition result, wherein the secondary recognition result includes the location of the data to be recognized category;

Slicing and dividing the segment to be identified according to the wifi connection time to obtain at least one wifi connection time segment, and marking the wifi connection time segment according to the location category of the data to be identified to obtain user location tagging information;

Generating the user track of the user according to the user location tag information, the original wifi data and the gps information.
The computer-readable storage medium according to claim 15, wherein the original wifi data includes wifi name data, and performing data preprocessing on the original wifi data to obtain the data to be identified includes:

Perform data cleaning processing on the wifi name data to obtain a data cleaning result;

Carry out word segmentation processing to wifi name data in the data cleaning result, obtain wifi word segmentation array;

The stop words in the wifi word segmentation array are removed to obtain the data to be recognized.
The computer-readable storage medium according to claim 16, wherein, performing word segmentation processing on the wifi name data in the data cleaning result, and obtaining the wifi word segmentation array includes:

Carry out word segmentation to the wifi name data in the data cleaning result, obtain sequence array;

Constructing the directed acyclic graph of the sequence array according to the preset prefix dictionary, and calculating the probability of each path in the directed acyclic graph respectively;

According to the path corresponding to the maximum probability in the directed acyclic graph, an optimal word segmentation result is obtained, and according to the optimal word segmentation result, the wifi name data in the data cleaning result is segmented to obtain a wifi word segmentation array.
The computer-readable storage medium according to any one of claims 15-17, wherein, performing a primary recognition on the data to be recognized according to a preset expert rule dictionary, and obtaining a recognition result includes:

Matching the data to be identified with the location words in the expert rule dictionary;

If the matching is successful, the location category corresponding to the location word that the data to be identified is successfully matched is used as a recognition result;

If the matching fails, set the first recognition result as recognition failure.
The computer-readable storage medium according to claim 18, wherein said inputting said to-be-recognized data into a pre-trained wifi recognition model to obtain a secondary recognition result comprises:

The data to be identified is input to the word vector layer in the wifi recognition model, and the data to be identified is converted into a word vector sequence;

The word vector is input to the maximum pooling layer in the wifi recognition model to obtain the maximum pooling result;

The maximum pooling result is input into the fully connected hidden layer in the wifi identification model, and the output result of the fully connected hidden layer is classified by the Softmax function to obtain the secondary identification result.
A method and device for identifying a user trajectory, wherein the method and device for identifying a user trajectory include:

Obtaining module, obtains original wifi data and gps information of user in the time period to be identified, and described original wifi data comprises wifi connection time;

A preprocessing module, configured to perform data preprocessing on the original wifi data to obtain data to be identified;

A primary recognition module, configured to perform primary recognition on the data to be recognized according to a preset expert rule dictionary to obtain a primary recognition result, and when the primary recognition result is a successful recognition, obtain the location category of the data to be recognized ;

A secondary recognition module, configured to input the data to be recognized into a pre-trained wifi recognition model when the primary recognition result is a recognition failure, to obtain a secondary recognition result, wherein the secondary recognition result includes the location category of the data to be identified;

A labeling module, configured to slice and divide the segment to be identified according to the wifi connection time to obtain at least one segment of the wifi connection time segment, and mark the wifi connection time segment according to the location category of the data to be identified to obtain User location marking information;

A trajectory drawing module, configured to generate the user trajectory of the user according to the user location label information, the original wifi data and the GPS information.