A site selection method, device and electronic equipment
By using federated learning and deep neural network technology in the user data server of telecom operators, combining commercial data and local telecom operation data to train the site selection scoring model, the unsatisfactory site selection results and privacy leakage in the existing site selection methods are solved. The problem is to achieve more efficient and accurate site selection results.
Patent Information
- Application Number
- CN202410075346.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-18
- Publication Date
- 2024-12-17
- Estimated Expiration
- 2044-01-18
AI Technical Summary
The existing site selection methods have problems such as unsatisfactory site selection results and privacy leakage, especially when using the site selection and scoring model of traditional machine learning algorithms, the scoring accuracy is low and a large amount of manual collection of sample data is required, resulting in high cost and low efficiency. .
Using federated learning technology to train a site selection scoring model in user data servers of multiple telecom operators, combine commercial data and local telecom operation data, and use deep neural networks to perform site selection scoring to ensure that the data is processed locally, avoid privacy leakage, and Improve scoring accuracy.
Through the combination of federated learning and deep neural network, the scoring accuracy of the site selection scoring model is improved, the site selection efficiency and effect are improved, and the privacy of user data is protected and privacy leakage is avoided.
Smart Images
Figure CN117952671B_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present invention relate to the field of artificial intelligence technology, and in particular to a site selection method, device and electronic equipment. Background Art
[0002] When choosing a site for supermarkets, shopping malls, cameras and other places or facilities that are sensitive to the flow of people, it is a very cautious matter. The correctness of the choice determines the future passenger flow and revenue benefits, so the site selection of these places or facilities requires extra caution.
[0003] At present, there are two commonly used site selection methods: one is manual site selection, which is based on manual observation and statistics to check the flow of people at multiple addresses, conduct sampling surveys, and then conduct input-output analysis to determine the best address; the other is site selection based on a site selection model, which is to collect sample data to train a site selection scoring model, input the relevant data of each candidate address into the site selection model, score each candidate address, and select the best address from the candidate addresses based on the score.
[0004] The first site selection method mentioned above mainly relies on the experience of researchers to select sites, which makes the site selection results subjective and uncertain. In addition, this method requires a lot of costs and has low site selection efficiency. The second site selection method also requires manual collection of sample data, so there are also problems with manual site selection. In addition, there is a risk of privacy leakage in the process of using sample data, and the site selection scoring model based on traditional machine learning algorithms has low scoring accuracy, resulting in unsatisfactory site selection results. Summary of the invention
[0005] The embodiments of the present invention provide a site selection method, device and electronic device, which are used to improve the site selection efficiency while solving the problems of privacy leakage and unsatisfactory site selection results in the existing site selection model site selection method.
[0006] The embodiment of the present invention adopts the following technical solutions:
[0007] In a first aspect, a site selection method is provided, which is applied to any user data server among user data servers of multiple telecommunications operators, and the method comprises:
[0008] Receiving commercial data of the candidate area sent by the central server;
[0009] Determining local telecommunication operation data of the candidate area;
[0010] Input the commercial data and the local telecommunication operation data into a local site selection scoring model to obtain a site selection score for the candidate area, wherein the site selection scoring model is obtained through federated learning of each user data server, and the sample data for training the site selection scoring model includes the local telecommunication operation data of the sample area and the commercial data of the sample area;
[0011] The deployment address of the target object is determined according to the site selection score of the candidate area.
[0012] In a second aspect, a site selection method is provided, which is applied to a central server, and the method includes:
[0013] The commercial data of the candidate area is sent to the user data servers of multiple telecom operators respectively, so that the user data servers input the commercial data and the local telecom operation data of the candidate area into the local site selection scoring model to obtain the site selection score of the candidate area, and determine the deployment address of the target object according to the site selection score of the candidate area, wherein the site selection scoring model is obtained by federated learning of the user data servers, and the sample data for training the site selection scoring model includes the local telecom operation data of the sample area and the commercial data of the sample area.
[0014] In a third aspect, a site selection device is provided, which is applied to any user data server among user data servers of multiple telecommunications operators, and the device includes:
[0015] A data receiving module, used for receiving the commercial data of the candidate area sent by the central server;
[0016] A data determination module, used to determine the local telecommunication operation data of the candidate area;
[0017] A site selection scoring module, used for inputting the commercial data and the local telecommunication operation data into a local site selection scoring model to obtain a site selection score for the candidate area, wherein the site selection scoring model is obtained by federated learning of each user data server, and the sample data for training the site selection scoring model includes the local telecommunication operation data of the sample area and the commercial data of the sample area;
[0018] A site selection module is used to determine the deployment address of the target object according to the site selection score of the candidate area.
[0019] In a fourth aspect, a site selection device is provided, which is applied to a central server, and the device includes:
[0020] The second data sending module is used to send the business data of the candidate area to the user data servers of multiple telecom operators respectively, so that the user data server inputs the business data and the local telecom operation data of the candidate area into the local site selection scoring model to obtain the site selection score of the candidate area, and determines the deployment address of the target object according to the site selection score of the candidate area, wherein the site selection scoring model is obtained by federated learning of each user data server, and the sample data for training the site selection scoring model includes the local telecom operation data of the sample area and the business data of the sample area.
[0021] In a fifth aspect, an electronic device is provided, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps of the method described in the first aspect or the second aspect are implemented.
[0022] In a sixth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in the first aspect or the second aspect are implemented.
[0023] At least one of the above technical solutions adopted in the embodiments of the present invention can achieve the following beneficial effects:
[0024] In the embodiment of the present application, on the one hand, user data servers of different telecom operators are trained by federated learning to obtain a site selection scoring model, thereby ensuring that the local data of each user data server does not go out, thereby avoiding privacy leakage; on the other hand, the site selection scoring model is trained by combining the commercial data of the candidate area and the local telecom operation data, which enriches the data volume of the site selection scoring model, thereby improving the scoring accuracy of the site selection scoring model and thus improving the site selection effect. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0026] Figure 1 It is a flowchart of a site selection method provided in one embodiment of the present application.
[0027] Figure 2 It is a schematic diagram of the structure of a deep neural network provided in one embodiment of the present application.
[0028] Figure 3 yes Figure 1 A detailed flow chart of step 102 in the site selection method is shown.
[0029] Figure 4 yes Figure 1 The figure shows an implementation architecture and data flow diagram of a site selection method.
[0030] Figure 5 It is a flowchart of a site selection method provided in another embodiment of the present application.
[0031] Figure 6 It is a schematic diagram of the implementation architecture and data flow of a site selection method provided in another embodiment of the present application.
[0032] Figure 7 It is a flowchart of a site selection method provided in yet another embodiment of the present application.
[0033] Figure 8 It is a flowchart of a site selection method provided in yet another embodiment of the present application.
[0034] Fig. 9 It is a structural schematic diagram of a site selection device provided in one embodiment of the present application.
[0035] Fig.10 It is a structural schematic diagram of a site selection device provided in another embodiment of the present application.
[0036] Fig.11 It is a structural schematic diagram of a site selection device provided in one embodiment of the present application.
[0037] Fig.12 It is a structural schematic diagram of a site selection device provided in another embodiment of the present application.
[0038] Fig.13 It is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION
[0039] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.
[0040] In order to improve the efficiency of site selection while solving the problems of privacy leakage and unsatisfactory site selection results in the existing site selection model site selection method, the embodiment of the present application provides a site selection method and device. The method and device can use federated learning technology to establish a site selection scoring model under the condition that the data of each telecom operator retains the local data server and does not go out, and jointly train the site selection scoring model with the commercial data of the candidate area and the local telecom operation data, which enriches the data volume of the site selection scoring model, and uses deep learning in the site selection scoring model to dig out deeper and more comprehensive features of the target area and improve the accuracy of site selection scoring.
[0041] The following is a description of a site selection method provided by the present application. The method may include two stages, the first stage is a stage of training a site selection scoring model, and the second stage is a stage of using the trained site selection scoring model to select a site. The following is a description of the first stage.
[0042] like Figure 1 As shown, in one embodiment, a site selection method provided by the present application can be applied to any user data server among user data servers of multiple telecommunications operators. The method may include:
[0043] Step 101, deploying an initial site selection scoring model locally.
[0044] In the embodiment of the present application, the site selection scoring model can be any scoring model. As an example, the site selection scoring model is a deep neural network.
[0045] Optionally, in order to make the site selection scoring model finally deployed on each user data server have the same scoring ability, the structure of the deep neural network deployed on each user data server is consistent. Specifically, the number of network layers, number of nodes, input format and output format of the deep neural network deployed on each user data server are the same. For example, the deep neural network deployed on each user data server is Figure 2 The structure shown.
[0046] Step 102, looping through the first designated step until the training termination condition of the site selection scoring model is met.
[0047] As an example, Figure 3 As shown, the first specifying step includes:
[0048] Step 1021: determine sample data, wherein the sample data includes: local telecommunication operation data of the sample area and business data of the sample area from the central server.
[0049] It can be seen that in the embodiment of the present application, the sample data for training the site selection scoring model includes two parts, one part is the local telecommunications operation data of the sample area, and the other part is the commercial data of the sample area from the central server.
[0050] The local telecommunications operation data of the sample area refers to the telecommunications operation data of the sample area stored in the user data server of a telecommunications operator. Such telecommunications operation data may include but is not limited to at least one of the following:
[0051] Customer traffic statistics;
[0052] Basic data of users;
[0053] The user's communication tariff data;
[0054] User call and network traffic data;
[0055] User’s social data;
[0056] User's surfing preference data.
[0057] It is understandable that telecom operators, as providers of Internet infrastructure, have unique inherent advantages and have access to massive amounts of data from massive users, such as real-name information data of users, service data (communication numbers and quantities, package information, communication charges, etc.), consumption data (call volume, network traffic, mobile payments, etc.), location data of different regions and time periods, social data, Internet preference data (browsing websites, commonly used apps, etc.), regional base station connection information, etc. Based on these data, a certain potential (such as consumption capacity) of a certain area can be reflected to a certain extent. Therefore, using these data as sample data for training the site selection scoring model can enable the trained site selection scoring model to identify these capabilities well, and then serve as the basis for whether to deploy the target object in the area, thereby improving the scoring accuracy.
[0058] As a specific example, the telecommunications operation data is the mobile signaling data of the sample area, including the average passenger flow during working hours on weekdays, the average passenger flow during rest time on weekdays, and the average passenger flow on holidays, such as the passenger flow from 8:00 to 18:00 from Monday to Friday, the passenger flow from 18:00 to 22:00 from Monday to Friday, and the passenger flow from 17:00 to 22:00 on Saturdays and Sundays; there are also gender and age distribution data of the passenger flow in the sample area, communication tariff data, call and network traffic data, social data, Internet preference data, etc.
[0059] The commercial data of the sample area may be provided by the central server. As an example, the commercial data may include but is not limited to at least one of the following:
[0060] Commercial housing rental data;
[0061] Statistical data of objects of the same type as the target object, for example, when the target object is a supermarket, the number of existing supermarkets in the sample area, etc.;
[0062] Business data of objects of the same type as the target object, for example, when the target object is a supermarket, the historical operating income of the existing supermarkets in the sample area.
[0063] It is understandable that the higher the rent of commercial housing in a certain area, the more supermarkets there are, and the higher the average income of supermarkets, the higher the location score of the area should be. Therefore, using the commercial data of the sample area as sample data to train the location scoring model can play a supervisory role, making the scoring accuracy of the trained location scoring model higher. For example, Figure 2 As shown, the dim sum operation data of the sample area is used as input, and the historical performance (historical business data) of the sample area is used as supervision to score the sample area.
[0064] Step 1022: input the sample data into the site selection scoring model to obtain the site selection score of the sample area.
[0065] It can be understood that when the sample data is input into the site selection scoring model, the site selection scoring model will accordingly output the site selection scoring result - the site selection score.
[0066] Step 1023: Determine the local gradient information of the site selection scoring model according to the site selection score of the sample area.
[0067] The specific process of calculating the local gradient information based on the site selection score output by the site selection scoring model belongs to the existing technology and will not be elaborated in this article.
[0068] Step 1024, encrypt the local gradient information and send it to the central server, so that the central server fuses the local gradient information of the sample area uploaded by each user data server to obtain fused gradient information and encrypts it and returns it.
[0069] Among them, each user data server can encrypt local gradient information through encryption technologies such as homomorphic encryption, differential privacy or secret sharing.
[0070] As an example, the central server can determine the weight of the local gradient information uploaded by each user data server according to the sample data contribution of each user data server when training the site selection scoring model; then, according to the weight of the local gradient information uploaded by each user data server, the local gradient information of the sample area uploaded by each user data server is fused to obtain fused gradient information. For example, there are 3 telecom operators in total, the first one provides 5 data, the second one provides 2 data, and the third one provides 3 data. Then, for the local gradient information A, B, and C transmitted by each party, 0.5*A+0.2*B+0.3*C is used as the final fused gradient information.
[0071] Of course, the central server may also determine the weight of the local gradient information uploaded by each user data server in other ways, and the embodiment of the present application does not limit this.
[0072] Step 1025, receiving the encrypted fused gradient information, and updating the parameters of the local site selection scoring model based on the fused gradient information.
[0073] The specific process of updating model parameters according to the fused gradient information also belongs to the prior art and will not be described in detail in this article.
[0074] After the above steps 1021 to 1025, each telecom operator obtains the same site selection scoring model locally, which enables each operator to have the same scoring ability when scoring candidate areas.
[0075] Figure 4 Shows Figure 1 The following is a schematic diagram of the implementation architecture and data flow of a site selection method. Figure 4 As shown in the figure, for the location selection of supermarkets as the target object, the subjects and processes involved in the location selection scoring model training process are as follows:
[0076] (1) The central server collects commercial data such as commercial housing rents and the number of supermarkets in the sample area, and uses the historical average operating income of multiple supermarkets in the sample area as the site selection score for the area. Generally speaking, the higher the average income, the higher the site selection score. The central server synchronizes the collected data to each telecom operator.
[0077] (2) The n telecom operators specify the telecom operation data to be provided according to the purpose and structure of the site selection scoring model to be constructed, and synchronize the telecom operation data and the business data collected in step (1) to their respective data servers.
[0078] (3) The n telecom operators respectively train the site selection scoring model according to the site selection scoring model to be constructed and the data obtained in step (1) and step (2), and calculate the local gradient information according to the output of the site selection scoring model.
[0079] (4) n telecom operators use encryption technologies such as homomorphic encryption, differential privacy, or secret sharing to encrypt local gradient information and send the encrypted local gradient information to the central server.
[0080] (5) The central server uses the data volume of each telecom operator as the weight, and performs weighted average of the local gradient information of each party as the fused gradient information. For example, there are three telecom operators in total. The first one provides 5 data, the second one provides 2 data, and the third one provides 3 data. Then, for the local gradient information A, B, and C transmitted by each party, 0.5*A+0.2*B+0.3*C is used as the fused gradient information.
[0081] (6) The central server sends the fused gradient information to each telecom operator. Each telecom operator decrypts the received fused gradient and uses the decrypted fused gradient information to update their own model parameters.
[0082] Repeat steps (3)-(6) until the agreed model training termination condition is reached.
[0083] After the above process, each (n) telecom operator obtains the same supermarket location scoring model locally, which enables each operator to have the same scoring ability when scoring candidate areas.
[0084] Figure 1 The illustrated embodiment provides a site selection method. When constructing a site selection scoring model, on the one hand, user data servers of different telecom operators are trained by federated learning to obtain a site selection scoring model, thereby ensuring that the local data of each user data server does not go out, thereby avoiding privacy leakage; on the other hand, the site selection scoring model is trained by combining the commercial data of the candidate area and the local telecom operation data, which enriches the data volume of the site selection scoring model, thereby improving the scoring accuracy of the site selection scoring model and thus improving the site selection effect.
[0085] also, Figure 1 The illustrated embodiment provides a site selection method that uses deep learning in a site selection scoring model, which can dig out deeper and more comprehensive features of each region, thereby improving the accuracy of site selection scoring.
[0086] The first node of a site selection method provided in an embodiment of the present application is described above, and the second node will be introduced next.
[0087] like Figure 5 As shown, another embodiment of the present application provides a site selection method, which is applied to any user data server among user data servers of multiple telecommunications operators. The method may include:
[0088] Step 103: receiving the commercial data of the candidate area sent by the central server.
[0089] As mentioned above, the commercial data of the candidate area may be provided by the central server. As an example, the commercial data may include but is not limited to at least one of the following:
[0090] Commercial housing rental data;
[0091] Statistical data of objects of the same type as the target object, for example, when the target object is a supermarket, the number of existing supermarkets in the sample area, etc.;
[0092] Business data of objects of the same type as the target object, for example, when the target object is a supermarket, the historical operating income of the existing supermarkets in the sample area.
[0093] Step 104: Determine local telecommunication operation data of the candidate area.
[0094] As mentioned above, the local telecommunications operation data of the candidate area refers to the telecommunications operation data of the candidate area stored in the user data server of a telecommunications operator. The telecommunications operation data may include but is not limited to at least one of the following:
[0095] Customer traffic statistics;
[0096] Basic data of users;
[0097] The user's communication tariff data;
[0098] User call and network traffic data;
[0099] User’s social data;
[0100] User's surfing preference data.
[0101] Step 105: Input the commercial data and the local telecommunication operation data into a local site selection scoring model to obtain a site selection score for the candidate area.
[0102] As described above, the site selection scoring model is obtained through federated learning of each user data server, and the sample data for training the site selection scoring model includes local telecommunication operation data of the sample area and commercial data of the sample area.
[0103] Step 106: Determine the deployment address of the target object according to the site selection score of the candidate area.
[0104] The target object may include but is not limited to the following:
[0105] Supermarkets;
[0106] Smart facilities in smart cities, such as smart manhole covers, smart trash cans or smart cameras, etc.
[0107] That is to say, the site selection method provided in the embodiment of the present application can be used not only for the site selection of supermarkets, but also for the site selection of important urban facilities in the urban operation management service platform, such as the site selection of smart manhole covers, smart trash cans, cameras, etc.
[0108] In addition, although a site selection method provided in the embodiment of the present application is aimed at the integrated use of data from multiple telecom operators, it can also be used for the secure sharing and operation of telecom operator data, providing ideas for the application of telecom operator data under the premise of legality and compliance, and can also be used for the secure sharing and operation of urban operation data. City managers use this technical solution to open urban operation data to social entities, which is conducive to improving the openness and intelligence of urban governance.
[0109] After scoring by the site selection scoring model, each telecom operator can achieve the purpose of outputting the site selection score of a certain input area. In practical applications, for a given batch of candidate areas, the site selection scoring model can be used to obtain the site selection score results of each area in the batch of candidate areas, sort the site selection scores of the batch of candidate areas, and select the candidate areas with the highest ranking as the target areas for deploying the target objects, so as to achieve the purpose of automatic site selection.
[0110] Figure 6 Shows Figure 5 The embodiment shown in the figure provides a schematic diagram of the implementation architecture and data flow of a site selection method. Figure 6 As shown in the figure, for the location selection of a supermarket, the location selection process is as follows:
[0111] (1) n telecom operators obtain commercial data of the target area (e.g., candidate area) from the central server, such as commercial housing rents, the number of supermarkets, etc. The data feature types and normalization methods must be consistent with those in the training phase;
[0112] (2) The n telecom operators each upload the data obtained in step (1) to their own user data servers in combination with the telecom operation data of the target area stored by them;
[0113] (3) n telecom operators input the business data from the central server and the local telecom operation data into the local site selection scoring model;
[0114] (4) The location scoring models corresponding to the n telecom operators output the location scores of the target areas respectively.
[0115] After scoring by the site selection scoring model, each telecom operator can achieve the purpose of outputting the site selection score of a certain input area. In practical applications, for a given batch of candidate areas, the site selection scoring model can be used to obtain the site selection score results of each area in the batch of candidate areas, sort the site selection scores of the batch of candidate areas, and select the candidate areas with the highest ranking as the target areas for deploying the target objects, so as to achieve the purpose of automatic site selection.
[0116] Optional, such as Figure 7 As shown, another embodiment of the present application further provides a site selection method, which can be applied to a central server. The method may include:
[0117] Step 701: Send the commercial data of the sample area to each user data server.
[0118] The commercial data of the sample area may be provided by the central server. As an example, the commercial data may include but is not limited to at least one of the following:
[0119] Commercial housing rental data;
[0120] Statistical data of objects of the same type as the target object, for example, when the target object is a supermarket, the number of existing supermarkets in the sample area, etc.;
[0121] Business data of objects of the same type as the target object, for example, when the target object is a supermarket, the historical operating income of the existing supermarkets in the sample area.
[0122] Step 702, looping through the second designated step until the training termination condition of the site selection scoring model is met.
[0123] Wherein, the second specifying step comprises:
[0124] Receive local gradient information encrypted and uploaded by each user data server, wherein the local gradient information is determined according to the site selection score of the sample area after the user number server inputs sample data into the site selection scoring model to obtain the site selection score of the sample area; the sample data includes: local telecommunication operation data of the sample area and commercial data of the sample area;
[0125] Fusing the local gradient information of the sample area uploaded by each user data server to obtain fused gradient information;
[0126] The fused gradient information is encrypted and sent to each user data server, so that each user data server updates the parameters of the local site selection scoring model based on the fused gradient information.
[0127] As an example, the fusing of the local gradient information of the sample area uploaded by each user data server to obtain the fused gradient information includes:
[0128] Determining the weight of the local gradient information uploaded by each user data server according to the sample data contribution of each user data server when training the site selection scoring model;
[0129] According to the weight of the local gradient information uploaded by each user data server, the local gradient information of the sample area uploaded by each user data server is fused to obtain fused gradient information.
[0130] Of course, the central server may also determine the weight of the local gradient information uploaded by each user data server in other ways, and the embodiment of the present application does not limit this.
[0131] The local telecommunications operation data of the sample area refers to the telecommunications operation data of the sample area stored in the user data server of a telecommunications operator. Such telecommunications operation data may include but is not limited to at least one of the following:
[0132] Customer traffic statistics;
[0133] Basic data of users;
[0134] The user's communication tariff data;
[0135] User call and network traffic data;
[0136] User’s social data;
[0137] User's surfing preference data.
[0138] Figure 7 The illustrated embodiment provides a site selection method. When constructing a site selection scoring model, on the one hand, user data servers of different telecom operators are trained by federated learning to obtain a site selection scoring model, thereby ensuring that the local data of each user data server does not go out, thereby avoiding privacy leakage; on the other hand, the site selection scoring model is trained by combining the commercial data of the candidate area and the local telecom operation data, which enriches the data volume of the site selection scoring model, thereby improving the scoring accuracy of the site selection scoring model and thus improving the site selection effect.
[0139] also, Figure 7The illustrated embodiment provides a site selection method that uses deep learning in a site selection scoring model, which can dig out deeper and more comprehensive features of each region, thereby improving the accuracy of site selection scoring.
[0140] Optional, such as Figure 7 As shown, another embodiment of the present application further provides a site selection method, which can be applied to a central server. The method may include:
[0141] Step 703, sending the business data of the candidate area to the user data servers of multiple telecom operators respectively, so that the user data server inputs the business data and the local telecom operation data of the candidate area into the local site selection scoring model to obtain the site selection score of the candidate area, and determines the deployment address of the target object according to the site selection score of the candidate area, wherein the site selection scoring model is obtained by federated learning of the user data servers, and the sample data for training the site selection scoring model includes the local telecom operation data of the sample area and the business data of the sample area.
[0142] After scoring by the site selection scoring model, each telecom operator can achieve the purpose of outputting the site selection score of a certain input area. In practical applications, for a given batch of candidate areas, the site selection scoring model can be used to obtain the site selection score results of each area in the batch of candidate areas, sort the site selection scores of the batch of candidate areas, and select the candidate areas with the highest ranking as the target areas for deploying the target objects, so as to achieve the purpose of automatic site selection.
[0143] In summary, the site selection method provided in the embodiment of the present application can achieve at least one of the following technical effects:
[0144] 1) Based on horizontal federated learning technology, the telecom operation data of multiple telecom operators are used to jointly establish a site selection scoring model without leaving the local area. This not only protects data privacy, but also expands the features of the traditional site selection scoring model and improves the site selection effect;
[0145] 2) A site selection scoring model based on a fully connected deep neural network calculates the scores of candidate areas to obtain a list of candidate areas, which has higher scoring accuracy and improves the site selection effect;
[0146] 3) While ensuring the data security of operators, the data value of telecom operators is brought into play;
[0147] 4) With the continuous development of society, a good site selection is crucial to the profitability of supermarkets and other target objects. This solution makes full use of the data of telecom operators to improve the accuracy and efficiency of site selection;
[0148] 5) In addition to telecommunications operation data, commercial data of the target area is also introduced to enrich the sample data, which makes the trained site selection scoring model more effective and more accurate, thereby improving the site selection effect.
[0149] The above describes a site selection method provided by an embodiment of the present application. Corresponding to the above method embodiment, an embodiment of the present application also provides a site selection device, which is described below.
[0150] Fig. 9 1 is a schematic diagram of a site selection device 900 provided by an embodiment of the present application. In a software implementation, the site selection device 900 can be applied to any user data server of multiple telecom operators, and the device 900 may include: a model deployment module 9001 and a model training module 9002.
[0151] The model deployment module 9001 is used to deploy the initial site selection scoring model locally.
[0152] In the embodiment of the present application, the site selection scoring model can be any scoring model. As an example, the site selection scoring model is a deep neural network.
[0153] Optionally, in order to ensure that the site selection scoring models finally deployed on each user data server have the same scoring capabilities, the structures of the deep neural networks deployed on each user data server are consistent.
[0154] The model training module 9002 is used to execute the first specified step in a loop until the training termination condition of the site selection scoring model is met.
[0155] Please refer to the above for the first designated step.
[0156] Fig. 9 The site selection device 900 shown can achieve Figure 1 The method of the method embodiment of the present invention can achieve the same technical effect, and the specific reference can be made to the above Figure 1 The introduction of the site selection method of the illustrated embodiment will not be repeated here.
[0157] Fig.10 1 is a schematic diagram of the structure of a site selection device 900 provided by an embodiment of the present application. In a software implementation, the site selection device 900 can be applied to any user data server of a plurality of telecommunications operators, and the device 900 may include: a data receiving module 903, a data determining module 904, a site selection scoring module 905 and a site selection module 906.
[0158] The data receiving module 903 is used to receive the commercial data of the candidate area sent by the central server.
[0159] The data determination module 904 is used to determine the local telecommunication operation data of the candidate area.
[0160] The site selection scoring module 905 is used to input the commercial data and the local telecommunication operation data into a local site selection scoring model to obtain the site selection score of the candidate area.
[0161] The site selection module 906 is used to determine the deployment address of the target object according to the site selection score of the candidate area.
[0162] After scoring by the site selection scoring model, each telecom operator can achieve the purpose of outputting the site selection score of a certain input area. In practical applications, for a given batch of candidate areas, the site selection scoring model can be used to obtain the site selection score results of each area in the batch of candidate areas, sort the site selection scores of the batch of candidate areas, and select the candidate areas with the highest ranking as the target areas for deploying the target objects, so as to achieve the purpose of automatic site selection.
[0163] Fig.10 The site selection device 900 shown can achieve Figure 5 The method of the method embodiment of the present invention can achieve the same technical effect, and the specific reference can be made to the above Figure 5 The site selection method of the embodiment shown is not described in detail here.
[0164] Fig.11 1 is a schematic diagram of a site selection device 1100 provided by an embodiment of the present application. In a software implementation, the site selection device 1100 can be applied to any user data server of a plurality of telecommunications operators, and the device 1100 may include: a first data sending module 1101 and a model training module 1102.
[0165] The first data sending module 1101 is used to send the commercial data of the sample area to each user data server.
[0166] The model training module 1102 is used to execute the second specified step in a loop until the training termination condition of the site selection scoring model is met.
[0167] For the content of the second designated step, please refer to the above and will not be repeated here.
[0168] Fig.11 The site selection device 1100 can achieve Figure 6 The method of the method embodiment of the present invention can achieve the same technical effect, and the specific reference can be made to the above Figure 7 The introduction of the site selection method of the illustrated embodiment will not be repeated here.
[0169] Fig.121 is a schematic diagram of a site selection device 1100 provided by an embodiment of the present application. In a software implementation, the site selection device 1100 can be applied to any user data server of a plurality of telecommunication operators, and the device 1100 can include: a second data sending module 1103 .
[0170] The second data sending module 1103 is used to send the business data of the candidate area to the user data servers of multiple telecom operators respectively, so that the user data server inputs the business data and the local telecom operation data of the candidate area into the local site selection scoring model to obtain the site selection score of the candidate area, and determines the deployment address of the target object according to the site selection score of the candidate area, wherein the site selection scoring model is obtained by federated learning of each user data server, and the sample data for training the site selection scoring model includes the local telecom operation data of the sample area and the business data of the sample area.
[0171] After scoring by the site selection scoring model, each telecom operator can achieve the purpose of outputting the site selection score of a certain input area. In practical applications, for a given batch of candidate areas, the site selection scoring model can be used to obtain the site selection score results of each area in the batch of candidate areas, sort the site selection scores of the batch of candidate areas, and select the candidate areas with the highest ranking as the target areas for deploying the target objects, so as to achieve the purpose of automatic site selection.
[0172] Fig.12 The site selection device 1100 can achieve Figure 8 The method of the method embodiment of the present invention can achieve the same technical effect, and the specific reference can be made to the above Figure 8 The introduction of the site selection method of the illustrated embodiment will not be repeated here.
[0173] In summary, the site selection device provided in the embodiment of the present application can achieve at least one of the following technical effects:
[0174] 1) Based on horizontal federated learning technology, the telecom operation data of multiple telecom operators are used to jointly establish a site selection scoring model without leaving the local area. This not only protects data privacy, but also expands the features of the traditional site selection scoring model and improves the site selection effect;
[0175] 2) A site selection scoring model based on a fully connected deep neural network calculates the scores of candidate areas to obtain a list of candidate areas, which has higher scoring accuracy and improves the site selection effect;
[0176] 3) While ensuring the data security of operators, the data value of telecom operators is brought into play;
[0177] 4) With the continuous development of society, a good site selection is crucial to the profitability of supermarkets and other target objects. This solution makes full use of the data of telecom operators to improve the accuracy and efficiency of site selection;
[0178] 5) In addition to telecommunications operation data, commercial data of the target area is also introduced to enrich the sample data, which makes the trained site selection scoring model more effective and more accurate, thereby improving the site selection effect.
[0179] Fig.13 This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Fig.13 At the hardware level, the electronic device includes a processor, and optionally also includes an internal bus, a network interface, and a memory. The memory may include a memory, such as a high-speed random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. Of course, the electronic device may also include hardware required for other services.
[0180] The processor, network interface and memory can be interconnected through an internal bus, which can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.13 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.
[0181] The memory is used to store the program. Specifically, the program may include a program code, and the program code includes a computer operation instruction. The memory may include a memory and a non-volatile memory, and provides instructions and data to the processor.
[0182] The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it, forming an address selection device at the logical level. The processor executes the program stored in the memory and is specifically used to perform the following operations:
[0183] Receiving commercial data of the candidate area sent by the central server;
[0184] Determining local telecommunication operation data of the candidate area;
[0185] Input the commercial data and the local telecommunication operation data into a local site selection scoring model to obtain a site selection score for the candidate area, wherein the site selection scoring model is obtained through federated learning of each user data server, and the sample data for training the site selection scoring model includes the local telecommunication operation data of the sample area and the commercial data of the sample area;
[0186] The deployment address of the target object is determined according to the site selection score of the candidate area.
[0187] Alternatively, the processor executes the program stored in the memory and is specifically configured to perform the following operations:
[0188] The commercial data of the candidate area is sent to the user data servers of multiple telecom operators respectively, so that the user data servers input the commercial data and the local telecom operation data of the candidate area into the local site selection scoring model to obtain the site selection score of the candidate area, and determine the deployment address of the target object according to the site selection score of the candidate area, wherein the site selection scoring model is obtained by federated learning of the user data servers, and the sample data for training the site selection scoring model includes the local telecom operation data of the sample area and the commercial data of the sample area.
[0189] The above application Figure 1 , Figure 5 , Figure 7 or Figure 8The site selection method disclosed in the illustrated embodiment can be applied to a processor or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in the processor or an instruction in software form. The above processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in one or more embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in one or more embodiments of the present application can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module can be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.
[0190] The electronic device may also perform Figure 1 , Figure 5 , Figure 7 or Figure 8 The site selection method is not described in detail in this application.
[0191] The embodiment of the present application also provides a computer-readable storage medium, which stores one or more programs, wherein the one or more programs include instructions, which, when executed by a portable electronic device including a plurality of application programs, enable the portable electronic device to execute Figure 3 or Figure 4 The method of the embodiment shown is specifically used to perform the following operations:
[0192] Receiving commercial data of the candidate area sent by the central server;
[0193] Determining local telecommunication operation data of the candidate area;
[0194] Input the commercial data and the local telecommunication operation data into a local site selection scoring model to obtain a site selection score for the candidate area, wherein the site selection scoring model is obtained through federated learning of each user data server, and the sample data for training the site selection scoring model includes the local telecommunication operation data of the sample area and the commercial data of the sample area;
[0195] The deployment address of the target object is determined according to the site selection score of the candidate area.
[0196] The embodiment of the present application also provides a computer-readable storage medium, which stores one or more programs, wherein the one or more programs include instructions, which, when executed by a portable electronic device including a plurality of application programs, enable the portable electronic device to execute Figure 1 or Figure 5 The method of the embodiment shown is specifically used to perform the following operations:
[0197] The commercial data of the candidate area is sent to the user data servers of multiple telecom operators respectively, so that the user data servers input the commercial data and the local telecom operation data of the candidate area into the local site selection scoring model to obtain the site selection score of the candidate area, and determine the deployment address of the target object according to the site selection score of the candidate area, wherein the site selection scoring model is obtained by federated learning of the user data servers, and the sample data for training the site selection scoring model includes the local telecom operation data of the sample area and the commercial data of the sample area.
[0198] The computer readable storage medium can also execute Figure 7 or Figure 8 The site selection method is not described in detail in this application.
[0199] Of course, in addition to software implementation methods, the electronic device of the present application does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc. That is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.
[0200] In short, the above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of the present application shall be included in the protection scope of one or more embodiments of the present application.
[0201] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
[0202] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0203] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0204] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
Claims
1. A site selection method, It is characterized in that Applied to any user data server among user data servers of multiple telecommunications operators, the method comprises: Receiving commercial data of the candidate area sent by the central server; Determining local telecommunication operation data of the candidate area; Input the commercial data and the local telecommunication operation data into a local site selection scoring model to obtain a site selection score for the candidate area, wherein the site selection scoring model is obtained through federated learning of each user data server, and the sample data for training the site selection scoring model includes the local telecommunication operation data of the sample area and the commercial data of the sample area; Determine the deployment address of the target object according to the site selection score of the candidate area; The local site selection scoring model of the user data server is obtained based on the following steps: Deploy the initial site selection scoring model locally; The first specified step is executed in a loop until a training termination condition of the site selection scoring model is met; Wherein, the first specifying step comprises: Determine sample data, wherein the sample data includes: local telecommunication operation data of the sample area and business data of the sample area from the central server; Inputting the sample data into the site selection scoring model to obtain the site selection score of the sample area; Determining local gradient information of the site selection scoring model according to the site selection score of the sample area; The local gradient information is encrypted and sent to the central server, so that the central server fuses the local gradient information of the sample area uploaded by each user data server to obtain fused gradient information and encrypts it and returns it; The encrypted fused gradient information is received, and the parameters of the local site selection scoring model are updated based on the fused gradient information.
2. The method according to claim 1, It is characterized in that The site selection scoring model is a deep neural network, and the structure of the deep neural network deployed on each user data server is consistent.
3. The method according to claim 1, It is characterized in that The business data includes at least one of the following: Commercial housing rental data; Statistical data of objects of the same type as the target object; Business data of objects of the same type as the target object.
4. The method according to claim 1, It is characterized in that The local telecommunications operation data includes at least one of the following: Customer traffic statistics; Basic data of users; The user's communication tariff data; User call and network traffic data; User’s social data; User's surfing preference data.
5. The method according to claim 1, It is characterized in that The target object includes one of the following: Supermarkets; Smart facilities for smart cities.
6. A site selection method, It is characterized in that Applied to a central server, the method comprises: Sending the commercial data of the candidate area to user data servers of multiple telecom operators respectively, so that the user data servers input the commercial data and the local telecom operation data of the candidate area into a local site selection scoring model to obtain the site selection score of the candidate area, and determine the deployment address of the target object according to the site selection score of the candidate area, wherein the site selection scoring model is obtained by federated learning of each user data server, and the sample data for training the site selection scoring model includes the local telecom operation data of the sample area and the commercial data of the sample area; The local site selection scoring model of the user data server is obtained based on the following steps: Sending business data of the sample area to each user data server; The second specified step is executed in a loop until a training termination condition of the site selection scoring model is met; Wherein, the second specifying step comprises: Receive local gradient information encrypted and uploaded by each user data server, wherein the local gradient information is determined by the user data server inputting sample data into the site selection scoring model to obtain the site selection score of the sample area, and then according to the site selection score of the sample area; the sample data includes: local telecommunication operation data of the sample area and commercial data of the sample area; Fusing the local gradient information of the sample area uploaded by each user data server to obtain fused gradient information; The fused gradient information is encrypted and sent to each user data server, so that each user data server updates the parameters of the local site selection scoring model based on the fused gradient information.
7. The method according to claim 6, It is characterized in that The fusing the local gradient information of the sample area uploaded by each user data server to obtain the fused gradient information includes: Determining the weight of the local gradient information uploaded by each user data server according to the sample data contribution of each user data server when training the site selection scoring model; According to the weight of the local gradient information uploaded by each user data server, the local gradient information of the sample area uploaded by each user data server is fused to obtain fused gradient information.
8. A site selection device, It is characterized in that Applicable to any user data server among user data servers of multiple telecommunications operators, the device comprises: A data receiving module, used for receiving the commercial data of the candidate area sent by the central server; A data determination module, used to determine the local telecommunication operation data of the candidate area; A site selection scoring model, used for inputting the commercial data and the local telecommunication operation data into a local site selection scoring model to obtain a site selection score for the candidate area, wherein the site selection scoring model is obtained by federated learning of each user data server, and the sample data for training the site selection scoring model includes the local telecommunication operation data of the sample area and the commercial data of the sample area; A site selection module, used to determine the deployment address of the target object according to the site selection score of the candidate area; The local site selection scoring model of the user data server is obtained based on the following steps: Deploy the initial site selection scoring model locally; The first specified step is executed in a loop until a training termination condition of the site selection scoring model is met; Wherein, the first specifying step comprises: Determine sample data, wherein the sample data includes: local telecommunication operation data of the sample area and business data of the sample area from the central server; Inputting the sample data into the site selection scoring model to obtain the site selection score of the sample area; Determining local gradient information of the site selection scoring model according to the site selection score of the sample area; The local gradient information is encrypted and sent to the central server, so that the central server fuses the local gradient information of the sample area uploaded by each user data server to obtain fused gradient information and encrypts it and returns it; The encrypted fused gradient information is received, and the parameters of the local site selection scoring model are updated based on the fused gradient information.
9. A site selection device, It is characterized in that Applied to a central server, the device comprises: A data sending module, used to send the commercial data of the candidate area to the user data servers of multiple telecom operators respectively, so that the user data server inputs the commercial data and the local telecom operation data of the candidate area into a local site selection scoring model to obtain the site selection score of the candidate area, and determines the deployment address of the target object according to the site selection score of the candidate area, wherein the site selection scoring model is obtained by federated learning of each user data server, and the sample data for training the site selection scoring model includes the local telecom operation data of the sample area and the commercial data of the sample area; The local site selection scoring model of the user data server is obtained based on the following steps: Sending business data of the sample area to each user data server; The second specified step is executed in a loop until a training termination condition of the site selection scoring model is met; Wherein, the second specifying step comprises: Receive the local gradient information encrypted and uploaded by each user data server, wherein the local gradient information is determined by the user data server inputting sample data into the site selection scoring model to obtain the site selection score of the sample area, and then according to the site selection score of the sample area; the sample data includes: Local telecommunication operation data of the sample area and business data of the sample area; Fusing the local gradient information of the sample area uploaded by each user data server to obtain fused gradient information; The fused gradient information is encrypted and sent to each user data server, so that each user data server updates the parameters of the local site selection scoring model based on the fused gradient information.