Method for constructing user location recognition model and method for user location recognition

Abstract

The present application relates to a method for constructing a user location recognition model and a method for user location recognition. The method for constructing a user location recognition model includes: sample user feature data, sample user real location data corresponding to the sample user feature data, test user feature data, and test user real location data corresponding to the test user feature data; training a preset neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model; inputting the test user feature data into the trained neural network model to obtain a user location recognition result output by the trained neural network model; screening the trained neural network model according to the user location recognition result and the test user real location data, and using the screened trained neural network model as a user location recognition model. The user location recognition model can realize high-precision recognition of the user location.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method, device, storage medium and computer equipment for constructing a user location recognition model, and a method, device, storage medium and computer equipment for user location recognition. Background Art

[0002] Authentication refers to determining whether a user has the authority to request a certain business or service. For example, in password-based authentication, authentication is successful when the password entered by the user is consistent with the pre-set password; in location-based authentication, authentication is successful when the user's location meets the preset conditions.

[0003] In the scenario of authenticating the user's geographic location, the user's real location is identified based on the geographic location reported by the user, and then subsequent response operations are performed based on the identification result. However, the traditional user location identification method has the problem of low location identification accuracy. Summary of the invention

[0004] Based on this, in order to solve the above technical problems, a method, device, storage medium and computer device for constructing a user location recognition model that improves the accuracy of location recognition, as well as a user location recognition method, device, storage medium and computer device are provided.

[0005] A method for constructing a user location recognition model, the method comprising:

[0006] Acquire sample user characteristic data, sample user real location data corresponding to the sample user characteristic data, test user characteristic data, and test user real location data corresponding to the test user characteristic data;

[0007] Training a preset neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model;

[0008] Inputting the test user feature data into the trained neural network model to obtain a user location recognition result output by the trained neural network model;

[0009] The trained neural network model is screened according to the user location recognition result and the real location data of the test user, and the screened trained neural network model is used as the user location recognition model.

[0010] A device for constructing a user location recognition model, the device comprising:

[0011] A data acquisition module, used to acquire sample user characteristic data, sample user real location data corresponding to the sample user characteristic data, test user characteristic data, and test user real location data corresponding to the test user characteristic data;

[0012] A model training module, used to train a preset neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model;

[0013] A model self-checking module, used for inputting the test user feature data into the trained neural network model to obtain a user position recognition result output by the trained neural network model;

[0014] The model determination module is used to screen the trained neural network model according to the user location recognition result and the real location data of the test user, and use the screened trained neural network model as the user location recognition model.

[0015] A computer-readable storage medium stores a computer program, which, when executed by a processor, causes the processor to perform the following steps:

[0016] Acquire sample user characteristic data, sample user real location data corresponding to the sample user characteristic data, test user characteristic data, and test user real location data corresponding to the test user characteristic data;

[0017] Training a preset neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model;

[0018] Inputting the test user feature data into the trained neural network model to obtain a user location recognition result output by the trained neural network model;

[0019] The trained neural network model is screened according to the user location recognition result and the real location data of the test user, and the screened trained neural network model is used as the user location recognition model.

[0020] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the following steps:

[0021] Acquire sample user characteristic data, sample user real location data corresponding to the sample user characteristic data, test user characteristic data, and test user real location data corresponding to the test user characteristic data;

[0022] Training a preset neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model;

[0023] Inputting the test user feature data into the trained neural network model to obtain a user location recognition result output by the trained neural network model;

[0024] The trained neural network model is screened according to the user location recognition result and the real location data of the test user, and the screened trained neural network model is used as the user location recognition model.

[0025] The method, device, storage medium and computer equipment for constructing the user location recognition model mentioned above train a preset neural network model through sample user feature data and sample user real location data to obtain a trained neural network model; input test user feature data into the trained neural network model to obtain a user location recognition result output by the trained neural network model; screen the trained neural network model according to the user location recognition result and the test user real location data, use the screened trained neural network model as the user location recognition model, obtain the neural network model through sample data training, and then perform model self-checking through test data, thereby constructing a user location recognition model, and the user location recognition model can be used to recognize the geographic location later, and the user location recognition model is trained by the neural network model through sample data and then obtained by screening through test data, so that it can support high-precision recognition of the user location.

[0026] A method for identifying a user position, the method comprising:

[0027] Receiving an authentication request initiated by a target user, wherein the authentication request carries characteristic data of the target user;

[0028] Reading a preset user location recognition model, wherein the user location recognition model is constructed by a user location recognition model construction method;

[0029] The target user feature data is input into the user location recognition model, and the target user location recognition result is obtained according to the output data of the user location recognition model.

[0030] A user position identification device, the device comprising:

[0031] A request receiving module, used to receive an authentication request initiated by a target user, wherein the authentication request carries characteristic data of the target user;

[0032] A model reading module, used to read a preset user location recognition model, wherein the user location recognition model is constructed by a user location recognition model construction method;

[0033] The data processing module is used to input the target user feature data into the user location recognition model, and obtain the target user location recognition result according to the output data of the user location recognition model.

[0034] A computer-readable storage medium stores a computer program, which, when executed by a processor, causes the processor to perform the following steps:

[0035] Receiving an authentication request initiated by a target user, wherein the authentication request carries characteristic data of the target user;

[0036] Reading a preset user location recognition model, wherein the user location recognition model is constructed by a user location recognition model construction method;

[0037] The target user feature data is input into the user location recognition model, and the target user location recognition result is obtained according to the output data of the user location recognition model.

[0038] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the following steps:

[0039] Receiving an authentication request initiated by a target user, wherein the authentication request carries characteristic data of the target user;

[0040] Reading a preset user location recognition model, wherein the user location recognition model is constructed by a user location recognition model construction method;

[0041] The target user feature data is input into the user location recognition model, and the target user location recognition result is obtained according to the output data of the user location recognition model.

[0042] The above-mentioned user location identification method, device, storage medium and computer equipment receive an authentication request initiated by a target user, and the authentication request carries the target user characteristic data; read a preset user location identification model, and the user location identification model is constructed by a user location identification model construction method; input the target user characteristic data into the user location identification model, and obtain the target user location identification result according to the output data of the user location identification model, and realize the identification of the geographic location through the user location identification model. The user location identification model is obtained by training a neural network model through sample data and then screening it through test data, thereby realizing high-precision identification of the user location. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 An application environment diagram of a method for constructing a user location identification model in one embodiment;

[0044] Figure 2 A schematic diagram of a flow chart of a method for constructing a user location identification model in an embodiment;

[0045] Figure 3 is a schematic diagram of a flow chart of a method for identifying a user location in an embodiment;

[0046] Figure 4 A schematic diagram of construction and application of a user location identification model in an embodiment;

[0047] Figure 5 A timing diagram of a computing cluster invoking a user location identification model for authentication in one embodiment;

[0048] Figure 6 A schematic diagram of a model management platform performing model management in one embodiment;

[0049] Figure 7 is a structural block diagram of a device for constructing a user location identification model in one embodiment;

[0050] Figure 8 is a structural block diagram of a user position identification device in one embodiment;

[0051] Fig. 9 FIG. 4 is a structural block diagram of a computer device in one embodiment. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0053] Figure 1 FIG. 1 is an application environment diagram of a method for constructing a user location recognition model in an embodiment. Figure 1, the method for constructing a user location recognition model is applied to a system for constructing a user location recognition model. The system for constructing a user location recognition model includes a mobile terminal 102 and a server 104, and the mobile terminal 102 is connected to the server 104 through a network. The user can send sample user feature data, sample user real location data corresponding to the sample user feature data, test user feature data, and test user real location data corresponding to the test user feature data to the server 104 through the mobile terminal 102. The server 104 trains a preset neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model; inputs the test user feature data into the trained neural network model to obtain a user location recognition result output by the trained neural network model; and screens the trained neural network model according to the user location recognition result and the test user real location data, and uses the screened trained neural network model as a user location recognition model. Among them, the mobile terminal 102 can be a mobile phone, a tablet computer, a notebook, a desktop computer, etc., and the server 104 can be implemented by an independent server or a server cluster composed of multiple servers. The above application scenario is illustrated by applying the method for constructing a user location identification model to a system including a mobile terminal and a server, and is implemented through the interaction between the mobile terminal and the server. It can be understood that the method for constructing a user location identification model can also be applied to mobile terminals and servers.

[0054] like Figure 2 As shown, in one embodiment, a method for constructing a user location recognition model is provided. This embodiment mainly applies this method to the above Figure 1 The server 104 in FIG. Figure 2 , the method specifically comprises the following steps:

[0055] S202, obtaining sample user characteristic data, sample user real location data corresponding to the sample user characteristic data, test user characteristic data, and test user real location data corresponding to the test user characteristic data.

[0056] User feature data refers to data used to characterize user attributes. For example, a user initiates an authentication request through a client to request permission for a certain business or service. When initiating an authentication request, the user reports the user feature data to the server synchronously through the client. Sample user feature data refers to user feature data used to train a model, and test user feature data refers to user feature data used to test a trained model. Sample user real location data is used to characterize the actual location attributes of the sample user, and test user real location data is used to characterize the actual location attributes of the test user, such as characterizing whether the sample user is a location simulation user or a store user, and whether the test user is a location simulation user or a store user. The real location data can be in the form of 0 / 1, that is, marked by 0 and 1, such as 0 for a store user and 1 for a location simulation user. Specifically, taking the merchant privilege scenario in the game as an example, a user who needs to obtain privileges initiates an authentication request to the server, and the authentication request carries user feature data, which can be latitude and longitude. The server determines whether the user is actually at the merchant's location based on the latitude and longitude reported by the user, and distributes privileges accordingly based on the judgment result. Some users use location simulation to modify the location of their mobile phones to bypass the system authentication logic, thereby obtaining privileges even when they are not at the merchant's location.

[0057] There are two main ways to obtain privileges through location simulation. Method one is location simulation + device information modification, and method two is location simulation + device information modification + network proxy. Among them, method one bypasses the verification of the project business logic by modifying the location positioning information and device information. For users who perform location simulation through method one, their network IP (Internet Protocol) information and IP common location are different from those of normal users who actually report information. Based on the above considerations, user characteristics include the number of times the IP is used in the merchant, the number of people and the corresponding number of devices, and the distance between the IP common location and the merchant. Users who use method two to perform location simulation, in addition to using location simulation software to modify location positioning information and using device information modification software to modify device information, also use network proxies to bypass the authentication business logic. The use of network proxies makes it more difficult to determine whether the IP information is abnormal, and the use of network proxies has diverse characteristics. The characteristics of several network proxy usage methods are shown in Table 1:

[0058] Table 1 Network proxy usage

[0059]

[0060] When a user initiates an authentication request to obtain privileges, the user feature data reported synchronously includes the user's latitude and longitude, WIFI name, connected BSSID (information used to uniquely mark the router), network IP information, and the historical usage of the above information in privileged items, such as the user's IP usage duration, the number of IPs used by the user, the distribution data of people using the BSSID under a certain merchant, the distribution data of people using the IP in a merchant, and the overall IP usage. Specifically, user characteristics may include the number of BSSID matching merchants, the number of times the BSSID appears in a merchant, IP matching merchant data, the number of times the IP is used under a merchant, the user IP positioning distance offset, the number of times the user's strong verification fails, the user's maximum movement speed on the day, the verification result, whether it is multiple accounts, whether it is rooted, whether it is location simulation, the number of merchants the user has recently reached, the number of times the device has failed to strong verification, the number of accounts used by the device, the number of account skins, the number of characters, the number of merchants the device has recently reached, the device's recent maximum movement speed, the number of IPs recently used by the user, the number of IPs recently used by the device, the number of days the IP is used, the number of people using the IP under the merchant, the number of devices using the IP under the merchant, the number of different IPs that have recently appeared under the merchant, the number of recent users of the merchant, the number of devices that have recently appeared at the merchant, etc. The acquired user characteristic data can be stored in the form of text for building a user location recognition model.

[0061] S204: Train a preset neural network model according to the sample user feature data and the sample user's real location data to obtain a trained neural network model.

[0062] For example, if a certain IP appears in multiple merchants across the country, it is very likely that the user is using the same network to tamper with the latitude and longitude to simulate a visit to the store. Take the user feature data including IP positioning distance, the number of merchants the IP has historically visited, and the number of merchants the device has historically visited as an example. For example, user a has been marked as a real store user, and user b has been marked as a location simulation user. User feature data A = [6000, 1, 1] for user a, and user feature data B = [50000, 10, 2] for location simulation user b. The preset neural network model is trained based on the user feature data and the marked information, so that the output of the neural network model f(A) = real store user, and f(B) = location simulation user as much as possible. f represents the role of the neural network model and can be regarded as a portrait of a real store user / location simulation user.

[0063] In one embodiment, a preset neural network model is trained according to sample user feature data and sample user real location data, and obtaining a trained neural network model includes: determining the neural network model corresponding to the sample user feature data according to the matching relationship between the features corresponding to the sample user feature data and the preset neural network model input parameters; obtaining the sample user feature data and the sample user real location data corresponding to each neural network model; training each neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model. User feature data corresponds to features of different dimensions, and different neural network models correspond to different input parameters. According to the matching relationship between user features and neural network model input parameters, for example, user features include feature 1, feature 2, feature 3, feature 4, and feature 5, the input parameters of neural network model one include feature 1 and feature 2, and the input parameters of neural network model two include feature 3, feature 4, and feature 5, the neural network model corresponding to the user feature data can be determined. After acquiring feature 1 data, feature 2 data, feature 3 data, feature 4 data, feature 5 data, and the real location data corresponding to each feature data, it can be determined based on the matching relationship that feature 1 data, feature 2 data, and the real location data corresponding to the two feature data correspond to neural network model 1, and feature 3 data, feature 4 data, feature 5 data, and the real location data corresponding to the three feature data correspond to neural network model 2. Neural network model 1 is trained based on feature 1 data, feature 2 data, and the real location data corresponding to the two feature data, and neural network model 2 is trained based on feature 3 data, feature 4 data, feature 5 data, and the real location data corresponding to the three feature data.

[0064] S206, inputting the test user feature data into the trained neural network model to obtain the user location recognition result output by the trained neural network model.

[0065] Taking user c as a test user as an example, it is known that user c is a location simulation user. The user feature data C = [80000, 4, 2] of user c is input into the trained neural network model to obtain the user location recognition result output by the trained neural network model, that is, whether user c is a location simulation user or a real in-store user.

[0066] S208, screening the trained neural network model according to the user location recognition result and the real location data of the test user, and using the screened trained neural network model as the user location recognition model.

[0067] As mentioned above, for example, the trained neural network model includes model 1, model 2 and model 3. The user feature data C of user c is input into model 1, model 2 and model 3 respectively to obtain the user location recognition results output by the model. Specifically, the user location recognition results output by model 1 are: user c is a location simulation user, the user location recognition results output by model 2 are: user c is a location simulation user, and the user location recognition results output by model 3 are: user c is a real in-store user. The real location data of user c indicates that it is a location simulation user. Therefore, model 3 is eliminated, and model 1 and model 2 are retained as user location recognition models.

[0068] The method for constructing the above-mentioned user location recognition model trains a preset neural network model through sample user feature data and sample user real location data to obtain a trained neural network model; inputs test user feature data into the trained neural network model to obtain a user location recognition result output by the trained neural network model; screens the trained neural network model according to the user location recognition result and the test user real location data, uses the screened trained neural network model as the user location recognition model, obtains the neural network model through sample data training, and then performs model self-checking through test data, thereby constructing a user location recognition model, and subsequently the user location recognition model can be used to realize the recognition of the geographic location, and the user location recognition model is trained by the neural network model through sample data and then obtained by screening through test data, so that it can support high-precision recognition of the user location.

[0069] In one embodiment, after the trained neural network model obtained by screening is used as the user location recognition model, it also includes: real-time monitoring and receiving the store user identification and the corresponding user location reported by the client; obtaining the user feature data corresponding to the store user identification, inputting the user feature data into the user location recognition model, and obtaining the user location recognition result output by the user location recognition model; when the user location recognition result is inconsistent with the user location, using the user feature data and the corresponding user location as model correction data; updating and calibrating the user location recognition model according to the model correction data. There may be misjudgments when performing location recognition through the user location recognition model. It can be determined whether a misjudgment has occurred through manually reported data. When a misjudgment occurs, the manually reported data corresponding to the misjudgment is used as model correction data to update and calibrate the user location recognition model to improve the accuracy of the user location recognition model.

[0070] like Figure 3 As shown, in one embodiment, a method for identifying a user location is provided, and the method specifically includes the following steps:

[0071] S302, receiving an authentication request initiated by a target user, where the authentication request carries characteristic data of the target user.

[0072] S304, reading a preset user location recognition model, where the user location recognition model is constructed by a user location recognition model construction method.

[0073] S306: Input the target user feature data into the user location recognition model, and obtain the target user location recognition result according to the output data of the user location recognition model.

[0074] The above-mentioned user location identification method receives an authentication request initiated by a target user, and the authentication request carries characteristic data of the target user; reads a preset user location identification model, and the user location identification model is constructed by a method for constructing a user location identification model; inputs the characteristic data of the target user into the user location identification model, and obtains a target user location identification result based on the output data of the user location identification model, and realizes the identification of the geographic location through the user location identification model. The user location identification model is obtained by training a neural network model through sample data and then screening it through test data, thereby realizing high-precision identification of the user location.

[0075] In one embodiment, the number of authentication requests and the number of user location identification models are multiple; the target user feature data is input into the user location identification model, and before obtaining the target user location identification result according to the output data of the user location identification model, it also includes: obtaining the load coefficient of each user location identification model; according to the load coefficient, the authentication request is assigned to different user location identification models for processing. The load coefficient is an indicator used to characterize the authentication request that the user location identification model can currently process. For example, the user location identification model includes model one, model two and model three. The load coefficient of model one is 30%, the load coefficient of model two is 20%, and the load coefficient of model three is 50%. When 100 authentication requests are obtained, the authentication requests are assigned to different user location identification models for processing according to the load coefficient, that is, model one processes 30 authentication requests, model two processes 20 authentication requests, and model three processes 50 authentication requests.

[0076] In one embodiment, obtaining the load factor of each user location identification model includes: obtaining the attributes of each user location identification model; and determining the load factor of each user location identification model according to the attributes of each user location identification model and the number of authentication requests. The attributes of the user location identification model include the user location identification model's own attributes and application attributes. The user location identification model's own attributes may specifically include the amount of computation of the user location identification model. The application attributes of the user location identification model may include whether the user location identification model is in an effective state, the total number of user location identification model verifications, the number of credible judgments, the number of suspicious judgments, the number of launderings, etc.

[0077] In one embodiment, the construction and application diagram of the user location recognition model is as follows: Figure 4 As shown in the figure, the dotted box on the left is a schematic diagram of the business logic of the authentication request in a certain scenario. The sample library is obtained from the business background in the figure, and the sample data is stored in the form of text and input into the training cluster built based on Tensorflow in batches. The positive and negative sample data are used for training the neural network model, and the trained neural network model is self-checked to generate a user location recognition model. The model library includes multiple user location recognition models. The generated user location recognition model is placed in the computing cluster built based on Tensorflow for use. When a new user initiates an authentication request, the authentication request carries user feature data, and the user feature data is transmitted to the computing cluster for calculation. The user description information is calculated by the user location recognition model, and the user is authenticated according to the calculated user description information.

[0078] In one embodiment, the timing diagram of the computing cluster calling the user location recognition model for authentication is as follows: Figure 5As shown, the user logs in to the game on the client, and the game client initiates a login request to the game backend. The game backend verifies the login request initiated by the game client and issues an available game token. At this time, in the model training cluster, the training sample data is placed in the model training cluster for model training. The game client sends the acquired game token to the game's built-in SDK (Software Development Kit), and the game SDK sends the user information and game token to the privileged backend, requesting the privileged backend to issue an available game privilege token. At this time, the trained neural network model is self-checked in the model training cluster. The privileged backend responds to the game privilege request reported by the SDK and pre-generates user-related authentication information. The privileged backend requests an available model from the model management platform, and the trained neural network model in the model training cluster is measured, and the model management platform loads the applicable model according to the measurement results. The privileged backend performs user environment portrait matching calculations based on the acquired user information and the neural network model fed back by the model management platform to obtain the calculation results. Synchronously, the available models generated by the model training cluster are published to the model management platform. The authentication results of the privileged background and the results of the merchant's intervention will be reported to the model management platform to realize data collection. On the one hand, the privileged background will send the generated privileged token to the SDK, and the game SDK will pass the obtained privileged token to the game client. On the other hand, the model management platform reports the collected data to the model training cluster for sample calibration. The game client sends the obtained privileged token to the game background for verification, and the game background sends the privileged token to the privileged background for verification. After the verification is passed, the game background directly issues privileges to the game client.

[0079] The model management platform uses artificial intelligence models, which have the characteristics of low transparency, especially the use of neural network models. The platform can effectively manage, verify and publish models, which plays an important role in the stable operation of the model management platform. The model management platform is equipped with corresponding modules to complete the verification, management, publishing, data collection, and routing functions. Through the synergy of each model, the effective management and safe switching of each model are realized, so that the model management platform can realize safe and stable switching among different artificial intelligence models. In this way, location simulation users can be intelligently identified, effectively reducing the negative impact of location simulation users on the project. In addition, it can also greatly reduce the time cost and money cost of merchants accessing privileged activities. The model management platform can safely protect the model and avoid the impact of inferior models on online authentication.

[0080] In one embodiment, a schematic diagram of a model management platform performing model management is shown as follows: Figure 6As shown in the figure, the neural network model is trained and generated in the training cluster, and the model that passes the model self-test will be loaded into the model management platform. Models with different recognition capabilities are managed uniformly in the model management platform. By adjusting the proportion of different models, the platform can dynamically adjust the recognition characteristics of the online authentication platform. At the same time, when releasing a new model, the model grayscale function can be realized based on the model management platform, and rollback can be achieved when necessary, allowing the online platform to upgrade seamlessly with security guarantees.

[0081] It should be understood that although Figure 2-3 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 2-3 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0082] Based on the same idea as the above method, Figure 7 A schematic diagram of the structure of a device for constructing a user location recognition model according to an embodiment is shown. This embodiment is described by taking the device being set in the server 104 as an example.

[0083] like Figure 7 As shown, the construction device of the user location recognition model in this embodiment includes:

[0084] The data acquisition module 702 is used to acquire sample user characteristic data, sample user real location data corresponding to the sample user characteristic data, test user characteristic data, and test user real location data corresponding to the test user characteristic data;

[0085] The model training module 704 is used to train the preset neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model;

[0086] The model self-checking module 706 is used to input the test user feature data into the trained neural network model to obtain the user location recognition result output by the trained neural network model;

[0087] The model determination module 708 is used to screen the trained neural network model according to the user location recognition result and the actual location data of the test user, and use the screened trained neural network model as the user location recognition model.

[0088] In one embodiment, the device for constructing a user location recognition model also includes a model correction module, which is used to monitor and receive in real time the store user identification and the corresponding user location reported by the client; obtain user feature data corresponding to the store user identification, input the user feature data into the user location recognition model, and obtain the user location recognition result output by the user location recognition model; when the user location recognition result is inconsistent with the user location, use the user feature data and the corresponding user location as model correction data; and update and calibrate the user location recognition model according to the model correction data.

[0089] In one embodiment, the model training module is also used to determine the neural network model corresponding to the sample user feature data based on the matching relationship between the features corresponding to the sample user feature data and the preset neural network model input parameters; obtain the sample user feature data and the sample user's real location data corresponding to each neural network model; and train each neural network model separately according to the sample user feature data and the sample user's real location data to obtain a trained neural network model.

[0090] Figure 8 A schematic diagram of the structure of a user position identification device according to an embodiment is shown. The user position identification device according to the embodiment includes:

[0091] The request receiving module 802 is used to receive an authentication request initiated by a target user, where the authentication request carries target user feature data;

[0092] A model reading module 804 is used to read a preset user location recognition model, where the user location recognition model is constructed by a user location recognition model construction method;

[0093] The data processing module 806 is used to input the target user feature data into the user location recognition model, and obtain the target user location recognition result according to the output data of the user location recognition model.

[0094] In one embodiment, the number of authentication requests and the number of user location identification models are multiple, and the user location identification device also includes an allocation processing module for obtaining the load coefficient of each user location identification model; according to the load coefficient, the authentication requests are allocated to different user location identification models for processing.

[0095] In one embodiment, the allocation processing module is further used to obtain the attributes of each user location identification model; and determine the load factor of each user location identification model according to the attributes of each user location identification model and the number of authentication requests.

[0096] Fig. 9The internal structure diagram of a computer device in one embodiment is shown. The computer device may specifically be Figure 1 The mobile terminal 102 or the server 104 in FIG. Fig. 9 As shown, the computer device includes a processor 902, a memory 904, a network interface 906, an input device 908, a display screen 910, a camera 912, a sound collection device 914 and a speaker 916 connected through a system bus. Among them, the memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor can implement a method for constructing a user location recognition model / a method for identifying a user location. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor can execute a method for constructing a user location recognition model / a method for identifying a user location. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device may be a touch layer covered on the display screen, or may be a key, trackball or touchpad provided on the housing of the computer device, or may be an external keyboard, touchpad or mouse, etc.

[0097] Those skilled in the art will understand that Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0098] In one embodiment, the user location recognition model construction device provided by the present application can be implemented in the form of a computer program, and the computer program can be used in Fig. 9 The computer device shown in the figure is run. The memory of the computer device can store various program modules constituting the construction device of the user location recognition model, such as: Figure 7 The data acquisition module, model training module, model self-checking module and model determination module shown in the figure. The computer program composed of each program module enables the processor to execute the steps in the method for constructing the user location recognition model of each embodiment of the present application described in this specification.

[0099] For example, Fig. 9 The computer device shown can be Figure 7The data acquisition module in the construction device of the user location recognition model shown in the figure executes to acquire the sample user feature data, the sample user real location data corresponding to the sample user feature data, the test user feature data and the test user real location data corresponding to the test user feature data. The computer device can execute, through the model training module, training of the preset neural network model according to the sample user feature data and the sample user real location data to obtain the trained neural network model. The computer device can execute, through the model self-checking module, inputting the test user feature data into the trained neural network model to obtain the user location recognition result output by the trained neural network model. The computer device can execute, through the model determination module, screening of the trained neural network model according to the user location recognition result and the test user real location data, and use the screened trained neural network model as the user location recognition model.

[0100] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for constructing a user location recognition model / the method for identifying a user location. The method for constructing a user location recognition model / the steps of the method for identifying a user location may be the steps of the method for constructing a user location recognition model / the method for identifying a user location in the above-mentioned embodiments.

[0101] In one embodiment, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the processor executes the steps of the method for constructing a user location recognition model / the method for identifying a user location. The method for constructing a user location recognition model / the method for identifying a user location may be the steps in the method for constructing a user location recognition model / the method for identifying a user location in the above-mentioned embodiments.

[0102] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0103] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0104] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for constructing a user location recognition model, the method comprising: Acquire sample user characteristic data, sample user real location data corresponding to the sample user characteristic data, test user characteristic data, and test user real location data corresponding to the test user characteristic data; User feature data includes IP location distance, number of merchants reached by IP history, and number of merchants reached by device history; the sample user real location data is used to characterize the actual location attribute of the sample user; the test user real location data is used to represent the actual location attribute of the test user; the actual location attribute is the location of the simulated user or the in-store user; Training a preset neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model; Inputting the test user feature data into the trained neural network model to obtain a user location recognition result output by the trained neural network model; the user location recognition result is a location simulation user or an in-store user; The trained neural network model is screened according to the user location recognition result and the real location data of the test user, and the screened trained neural network model is used as the user location recognition model.

2. The method according to claim 1, characterized in that After using the trained neural network model obtained through screening as the user location recognition model, the method further includes: Real-time monitoring and receiving the store user ID and corresponding user location reported by the client; Acquire user feature data corresponding to the store-visiting user identifier, input the user feature data into the user location recognition model, and obtain a user location recognition result output by the user location recognition model; When the user location recognition result is inconsistent with the user location, using the user feature data and the corresponding user location as model correction data; The user position recognition model is updated and calibrated according to the model correction data.

3. The method according to claim 1, characterized in that The step of training a preset neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model includes: Determine the neural network model corresponding to the sample user feature data according to the matching relationship between the features corresponding to the sample user feature data and the preset neural network model input parameters; Obtain sample user feature data and sample user real location data corresponding to each neural network model; According to the sample user feature data and the sample user real location data, each neural network model is trained respectively to obtain a trained neural network model.

4. A method for identifying a user location, characterized in that: The method comprises: Receiving an authentication request initiated by a target user, wherein the authentication request carries characteristic data of the target user; Reading a preset user location recognition model, wherein the user location recognition model is constructed by the method described in any one of claims 1 to 3; The target user feature data is input into the user location recognition model, and the target user location recognition result is obtained according to the output data of the user location recognition model.

5. The method according to claim 4, characterized in that The number of the authentication requests and the number of the user location identification models are multiple; Before inputting the target user feature data into the user location recognition model and obtaining the target user location recognition result according to the output data of the user location recognition model, the method further includes: Obtaining the load factor of each user location identification model; According to the load factor, the authentication request is distributed to different user location identification models for processing.

6. The method according to claim 5, characterized in that The step of obtaining the load factor of each user location identification model includes: Get the properties of each user location recognition model; The load factor of each user location identification model is determined according to the attribute of each user location identification model and the number of the authentication requests.

7. A device for constructing a user location recognition model, characterized in that: The device comprises: A data acquisition module is used to acquire sample user characteristic data, sample user real location data corresponding to the sample user characteristic data, test user characteristic data, and test user real location data corresponding to the test user characteristic data; user characteristic data includes IP positioning distance, number of merchants reached by IP history, and number of merchants reached by device history; the sample user real location data is used to characterize the actual location attribute of the sample user; the test user real location data is used to represent the actual location attribute of the test user; the actual location attribute is the location of a simulated user or an in-store user; A model training module, used to train a preset neural network model according to the sample user feature data and the sample user real location data to obtain a trained neural network model; A model self-checking module, used for inputting the test user feature data into the trained neural network model to obtain a user location recognition result output by the trained neural network model; the user location recognition result is a location simulation user or an in-store user; The model determination module is used to screen the trained neural network model according to the user location recognition result and the real location data of the test user, and use the screened trained neural network model as the user location recognition model.

8. The device according to claim 7, characterized in that The device also includes a model correction module, which is used to monitor in real time and receive the store user identification and the corresponding user location reported by the client, obtain user feature data corresponding to the store user identification, input the user feature data into the user location recognition model, and obtain the user location recognition result output by the user location recognition model; when the user location recognition result is inconsistent with the user location, use the user feature data and the corresponding user location as model correction data, and update and calibrate the user location recognition model according to the model correction data.

9. The device according to claim 7, characterized in that The model training module is also used to determine the neural network model corresponding to the sample user feature data based on the matching relationship between the features corresponding to the sample user feature data and the preset neural network model input parameters, obtain the sample user feature data and the sample user's real location data corresponding to each neural network model, and train each neural network model according to the sample user feature data and the sample user's real location data to obtain a trained neural network model.

10. A user location identification device, characterized in that: The device comprises: A request receiving module, used to receive an authentication request initiated by a target user, wherein the authentication request carries characteristic data of the target user; A model reading module, used to read a preset user location recognition model, wherein the user location recognition model is constructed by the method described in any one of claims 1 to 3; The data processing module is used to input the target user feature data into the user location recognition model, and obtain the target user location recognition result according to the output data of the user location recognition model.

11. The device according to claim 10, characterized in that The number of the authentication requests and the number of the user location identification models are multiple, and the device also includes an allocation processing module, which is used to obtain the load coefficient of each user location identification model and allocate the authentication requests to different user location identification models for processing according to the load coefficient.

12. The device according to claim 11, characterized in that The allocation processing module is further used to obtain the attributes of each user location identification model, and determine the load factor of each user location identification model according to the attributes of each user location identification model and the number of authentication requests.

13. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the method according to any one of claims 1 to 6.

14. A computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 6.

Images