A Secure Personalized Travel Recommendation System Based on Federated Learning and Regional Avoidance

Abstract

The present invention discloses a secure personalized travel recommendation system based on federated learning and regional avoidance. The system includes a user terminal and a server. The user terminal first performs federated training on the global model, and the server aggregates the global model without user information through federated learning and then distributes it to the user terminal. The user terminal uses the globally aggregated model through federated learning to predict the user's personalized travel mode, and determines the travel route using the latest regional risk information and the path planning algorithm distributed by the server. The present invention trains the global model locally, which can achieve not leaking the user's personal data to the server, protecting the user's privacy and security, and can select the travel route according to the path planning algorithm distributed by the server, making reasonable use of the scheduling ability of the server. The present invention can be widely applied to the field of path planning.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and in particular to a secure personalized travel recommendation system based on federated learning and area avoidance. Background Art

[0002] Existing travel recommendation solutions can collect and summarize public information to provide users with risk information and travel restrictions in relevant areas. However, most solutions are limited to the active avoidance of users and it is difficult to ensure meeting the personalized needs of users. Also, since existing travel recommendation solutions cannot obtain the personal privacy data of users, existing path planning only considers factors such as the shortest path, the fewest turns, or the lowest price. Therefore, the recommended routes obtained are not the safest routes. In summary, the existing travel recommendation solutions have the following problems: The existing travel recommendation solutions transform the task of risk avoidance during travel into the active avoidance of users, which not only loses the scheduling ability of the central server for user travel but also cannot provide personalized services to users while ensuring the personal privacy security of users.

[0003] In view of the above problems, a personalized travel recommendation solution that can make full use of the scheduling ability of the central server and ensure the personal privacy security of users has important research significance. Summary of the Invention

[0004] In view of this, an embodiment of the present invention provides a secure personalized travel recommendation system based on federated learning and area avoidance that can make full use of the scheduling ability of the central server and ensure the personal privacy security of users.

[0005] An aspect of an embodiment of the present invention provides a secure personalized travel recommendation system based on federated learning and area avoidance, including a user terminal and a server;

[0006] The user terminal is configured to:

[0007] Obtain area risk information from the server and obtain a destination determined by the user according to the area risk information;

[0008] Obtain a first global model from the server, perform federated learning on the first global model according to the area risk information and the destination, and upload the first global model after federated learning training to the server, and use the first global model after federated learning training as a second global model;

[0009] Predict the travel mode of the user by using a third global model and in combination with the area risk information, and predict a first travel route of the user from a specified starting point to the destination by using a path planning algorithm and in combination with the travel mode;

[0010] The server is configured to:

[0011] Perform federated learning aggregation on the second global model, verify the second global model after federated learning aggregation on the public dataset, and send the second global model that meets the preset verification requirements and the path planning algorithm to the user terminal, and use the second global model that meets the preset verification requirements as the third global model.

[0012] Optionally,

[0013] The user terminal is further configured to upload the starting point, the destination, and the first travel route to the server, so that the server can update the regional risk information according to the starting point, the destination, and the first travel route.

[0014] Optionally,

[0015] The user terminal includes:

[0016] A travel information expansion unit, configured to expand at least one of the starting point, the destination, or the first travel route outward by a set range;

[0017] A travel information upload unit, configured to upload at least one of the starting point, the destination, or the first travel route after the expansion range to the server.

[0018] Optionally,

[0019] The user terminal includes:

[0020] A travel route prediction unit, configured to use the path planning algorithm according to a preset path calculation formula and combine the travel mode to predict the travel route of the user from the starting point to the destination;

[0021] The path calculation formula is:

[0022] Cost road = Length road + Abs(Risk user - Risk road ) × α × β

[0023] Wherein, Cost road represents the cost of the candidate path, Length road represents the length of the candidate path, Abs() represents the absolute value function, Risk user represents the virtual risk score of the user, Risk roaddenote the risk scores of the candidate paths determined according to the regional risk information, α denotes the risk concern coefficient that can be set by the user, and β denotes whether the user uses a preset target travel mode.

[0024] Optionally,

[0025] The server is further configured to determine candidate road network information according to the starting point and the destination, and send the candidate road network information to the user terminal for the user to determine a second travel route according to the candidate road network information.

[0026] Optionally,

[0027] The server is further configured to send a second global model that does not meet the preset inspection requirements to the user terminal for the user terminal to perform federated learning on the second global model that does not meet the preset inspection requirements.

[0028] One aspect of the embodiments of the present invention provides a secure personalized travel recommendation method based on federated learning and regional avoidance, including:

[0029] The user terminal obtains regional risk information from the server and obtains the destination determined by the user according to the regional risk information;

[0030] The user terminal obtains a first global model from the server, performs federated learning on the first global model according to the regional risk information and the destination, and uploads the first global model after federated learning training to the server. The first global model after federated learning training is used as a second global model;

[0031] The server performs federated learning aggregation on the second global model, inspects the second global model after federated learning aggregation on a public dataset, and sends the second global model that meets the preset inspection requirements and a path planning algorithm to the user terminal. The second global model that meets the preset inspection requirements is used as a third global model;

[0032] The user terminal uses the third global model and combines the regional risk information to predict the travel mode of the user, and uses the path planning algorithm and combines the travel mode to predict the first travel route of the user from the specified starting point to the destination.

[0033] Optionally, it further includes:

[0034] The user terminal uploads the starting point, the destination, and the first travel route to the server for the server to update the regional risk information according to the starting point, the destination, and the first travel route.

[0035] Optionally, the uploading of the starting point, the destination, and the first travel route to the server by the user terminal includes:

[0036] Expanding, by the user terminal, at least one of the starting point, the destination, or the first travel route by a set range;

[0037] Uploading, by the user terminal, at least one of the starting point, the destination, or the first travel route after the expansion of the range to the server.

[0038] Optionally, the predicting of the first travel route of the user from the specified starting point to the destination by using the path planning algorithm and combining with the travel mode includes:

[0039] Predicting, by the user terminal, the first travel route of the user from the starting point to the destination by using the path planning algorithm according to a preset path calculation formula and combining with the travel mode;

[0040] The path calculation formula is:

[0041] Cost road = Length road + Abs(Risk user - Risk road ) × α × β

[0042] Wherein, Cost road represents the cost of the candidate path, Length road represents the length of the candidate path, Abs() represents the absolute value function, Risk user represents the virtual risk score of the user, Risk road represents the risk score of the candidate path determined according to the regional risk information, α represents the risk concern coefficient that can be set by the user, and β represents whether the user uses the preset target travel mode.

[0043] Another aspect of the embodiments of the present invention further provides an electronic device, including a processor and a memory;

[0044] The memory is used for storing a program;

[0045] The processor executes the program to implement the above method.

[0046] Another aspect of the embodiments of the present invention further provides a computer-readable storage medium, where the storage medium stores a program, and the program is executed by a processor to implement the above method.

[0047] An embodiment of the present invention also discloses a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the foregoing method.

[0048] The system of the present invention includes a user terminal and a server. The user terminal can obtain regional risk information from the server and a destination determined by the user according to the regional risk information. The user terminal can obtain a first global model from the server and perform federated learning training using local user information. Compared with uploading user information to the server and allowing the server to train the first global model, the present invention can ensure that user information does not flow out of the user terminal and protect the personal privacy and security of users. The first global model after federated learning on the user terminal is used as the second global model. The server can perform federated learning aggregation on the second global model, verify the second global model after federated learning aggregation on a public dataset, and send the second global model that meets the preset verification requirements and the path planning algorithm to the user terminal. By uploading the second model that has nothing to do with user information after training on the user terminal, the server can integrate the global model information of different users to obtain a better global model, and then re-send it to the user for continuous training. Repeating this process, a global model that ensures the non-disclosure of user privacy information and has practicality can be obtained. The second global model that meets the preset verification requirements is used as the third global model. The user terminal can use the third global model and combine the regional risk information to predict the user's travel mode, and use the path planning algorithm and combine the travel mode to predict the user's first travel route. Based on the travel mode predicted by the global model, the path planning algorithm sent by the server, and the latest regional risk information, the scheduling ability of the server can be integrated to provide a travel route that meets the personalized requirements and travel requirements of the user, realizing safe, efficient and accurate travel recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0050] Figure 1 It is a system framework diagram of a secure personalized travel recommendation system based on federated learning and regional avoidance provided by an embodiment of the present invention;

[0051] Figure 2Scenario example diagram of a secure personalized travel recommendation system based on federated learning and area avoidance provided by an embodiment of the present invention;

[0052] Figure 3 Scenario example diagram of a secure personalized travel recommendation system based on federated learning and area avoidance provided by an embodiment of the present invention;

[0053] Figure 4 Scenario example diagram of a secure personalized travel recommendation system based on federated learning and area avoidance provided by an embodiment of the present invention;

[0054] Figure 5 Flow schematic diagram of a secure personalized travel recommendation method based on federated learning and area avoidance provided by an embodiment of the present invention. Detailed implementation manners

[0055] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0056] Refer to Figure 1 An embodiment of the present invention provides a secure personalized travel recommendation system based on federated learning and area avoidance. The system includes a user terminal 10 and a server 20. Among them, the user terminal 10 can be specifically used for the following processes:

[0057] S100: Obtain area risk information from the server, and obtain a destination determined by the user according to the area risk information.

[0058] Specifically, the area risk information may include the latest risk information of different areas currently. Examples are: risk information such as whether a fire has occurred, whether an earthquake has occurred, and traffic jams. In order to provide a secure travel recommendation plan for users, areas where risks are currently occurring or predicted to occur soon can be marked to remind users to avoid risk areas when choosing a destination.

[0059] S110: Obtain a first global model from the server, perform federated learning on the first global model according to the area risk information and the destination, and upload the first global model after federated learning training to the server, and use the first global model after federated learning training as a second global model.

[0060] Specifically, the first global model can be trained through federated learning locally on the user terminal. The training process can use the personal information of the user stored locally on the user terminal. After training locally on the user terminal and meeting the preset requirements, the first global model without user information can be uploaded to the server, avoiding the leakage of the user's personal information to the server.

[0061] S120: Use the third global model and combine the regional risk information to predict the user's travel mode, and use the path planning algorithm and combine the travel mode to predict the first travel route of the user from the specified starting point to the destination.

[0062] Specifically, the third global model that passes the inspection can avoid risks according to the user's personalized needs and combine the latest regional risk information to predict the user's travel mode. The travel mode can include: public transportation, high-speed rail, airplane, cycling, walking, etc. Then, based on the path planning algorithm and the travel mode, the first travel route that the user may choose can be predicted.

[0063] The server 20 can be specifically used for the following process:

[0064] S130: Perform federated learning aggregation on the second global model, inspect the second global model after federated learning aggregation on the public dataset, and send the second global model that meets the preset inspection requirements and the path planning algorithm to the user terminal, and use the second global model that meets the preset inspection requirements as the third global model.

[0065] Specifically, after local training on the user terminal, the second global model irrelevant to user information is uploaded. The server can integrate the second global model information of different users to obtain a better global model, and send the third global model obtained after integration and inspection to the user terminal for continued training.

[0066] In addition, the server can also send the second global model that does not meet the preset inspection requirements to the user terminal for the user terminal to perform federated learning on the second global model that does not meet the preset inspection requirements again and wait for the opportunity to upload the global model next time.

[0067] After the user determines the first travel route, the present invention can also upload the starting point, the destination, and the first travel route to the server through the user terminal for the server to update the regional risk information according to the starting point, the destination, and the first travel route. Specifically, it can include the following steps:

[0068] S1: Expand at least one of the starting point, the destination, or the first travel route outward by a set range through the user terminal.

[0069] Specifically, during the process of the user providing a partial path to the server, the user's identity needs to be strictly protected from being inferred. Therefore, to protect user privacy, a buffer zone can be formed by expanding the range set outward from the user's different starting points to different destinations and through different regions, ensuring that the user's specific travel information is not leaked to the server.

[0070] According to the latest regional risk information, regions where risk situations are likely to occur are recorded as high-risk regions, and regions where risk situations are not likely to occur are recorded as low-risk regions.

[0071] Case 1: When the user starts from the first high-risk region, passes through a low-risk region, and then enters the second high-risk region, the path passing through the low-risk region can be expanded outward by a set range to form a buffer zone, so that the path passed by the user becomes a regional range rather than a definite path. Furthermore, the buffer zone can be marked as a high-risk region and uploaded to the server to update the regional risk information in the server.

[0072] Specifically, a specific example of Case 1 can refer to Figure 2 , Figure 2 where the user starts from a starting point in the high-risk region h1, passes through the low-risk region l, and arrives at a destination in the high-risk region h2. At this time, the path passing through the low-risk region l can be expanded outward by a set range to form a buffer zone lexpand, and the buffer zone lexpand can be marked as a high-risk region and uploaded to the server to update the regional risk information.

[0073] Case 2: When the user starts from a high-risk region, passes through a low-risk region, and arrives at a low-risk destination, the path passing through the low-risk region can be expanded outward by a set range to form a buffer zone, and a buffer zone can be formed by expanding the range outward with the destination as the center, so that the path passed by the user becomes a regional range rather than a definite path, and the destination is also a buffer zone. Furthermore, each buffer zone can be marked as a high-risk region and uploaded to the server to update the regional risk information in the server.

[0074] Specifically, a specific example of Case 2 can refer to Figure 3 , Figure 3 where the user starts from a starting point in the high-risk region h, passes through the low-risk region l, and arrives at a destination in the low-risk region l. At this time, the path passing through the low-risk region l and the destination can be expanded outward by a set range to form a buffer zone. In this specific example, a circular buffer zone with a set radius can be formed outward with the destination as the center. Furthermore, the buffer zone can be marked as a high-risk region and uploaded to the server to update the regional risk information.

[0075] Case 3: When the user departs from a high-risk area, passes through a low-risk area, and then enters the high-risk area again, the path passing through the low-risk area can be expanded outward by a set range to form a buffer zone, and the starting point and the destination can be expanded outward by a set range to form a buffer zone, so that the path passed by the user becomes an area range rather than a definite path. At the same time, the buffer zones formed by the outward buffering of the destination and the starting point cannot be searched, and then the buffer zones can be marked as high-risk areas and uploaded to the server to update the regional risk information in the server.

[0076] Specifically, a specific example of Case 3 can be referred to Figure 4 , Figure 4 In, the user departs from the starting point located in the high-risk area h, passes through the low-risk area l, and arrives at the destination located in the high-risk area h. At this time, the path passing through the low-risk area l can be expanded outward by a set range to form a buffer zone, and taking the starting point and the destination as the centers respectively, circular buffer zones with a set radius are expanded outward. The buffer zones are marked as high-risk areas and uploaded to the server to update the regional risk information.

[0077] S2. Upload at least one of the expanded starting point, the destination, or the first travel route to the server through the user terminal.

[0078] In some embodiments of the present invention, the above S120 is introduced, which is the process of predicting the first travel route of the user from the specified starting point to the destination by using the path planning algorithm and combining the travel mode. Next, this process will be further described.

[0079] Specifically, through the user terminal, using the path planning algorithm according to the preset path calculation formula, and combining the travel mode to predict the first travel route of the user from the starting point to the destination;

[0080] The preset path calculation formula is:

[0081] Cost road = Length road + Abs(Risk user - Risk road ) × α × β

[0082] Among them, Cost road represents the cost of the candidate path, Length road represents the length of the candidate path, Abs() represents the absolute value function, Risk user represents the virtual risk score of the user, Risk roadIt represents the risk score of the candidate path determined according to the regional risk information, α represents the risk concern coefficient that can be set by the user, and β represents whether the user uses the preset target travel mode.

[0083] Among them, the initial value of α can be set to 10, and the calculation formula of β is:

[0084]

[0085] V user It represents the travel mode of the user's current trip, and V high represents the preset target travel mode. The target travel mode may be that the user has a relatively high probability of encountering a risk situation when choosing this travel mode. In addition, the user's virtual risk score Risk user can be determined according to the risk situation of the area where the user is currently located.

[0086] In addition to using the path planning algorithm to determine the first travel route, the server of the present invention can also send the candidate road network information to the user terminal for the user to determine the second travel route according to the candidate road network information.

[0087] Specifically, in the system of the present invention, in order to protect the user's personal privacy, all operations involving the user's personal privacy information will be performed on the user terminal, such as personalized travel mode prediction and path planning. In order to make full use of the scheduling ability of the server, the server can provide the candidate road network information required by the user terminal according to the latest regional risk information and in combination with the user's starting point and destination, reducing the complexity of path planning and the storage space required by the user terminal and other conditions.

[0088] Among them, the above-mentioned returned candidate road network information may specifically include: route nodes, route information (whether there is a route between nodes, route length, etc.), route additional information (whether it is impassable due to maintenance and other conditions), etc., as well as other optional information.

[0089] The system of the present invention first forms the preliminary regional risk attributes through the latest risk information of each region, and then performs subsequent risk updates in a privacy protection manner; at the same time, it uses the federated learning framework to break the data barrier, and quickly restores the accuracy of the personalized travel mode recommendation service by using data from multiple parties, and effectively predicts the user-level travel preferences during the possible risk period; finally, it uses the risk-aware path planning method to provide the user with the safest and user-preferred best path. The present invention provides a safe and reliable personalized travel plan in a privacy protection manner, realizing the accuracy of travel mode prediction for avoiding regional risks and the safety of the best travel route.

[0090] Next, a specific example will be used to illustrate the process of the present invention for realizing travel recommendation.

[0091] S1: The user terminal obtains the latest regional risk information from the server to the local, and then the latest regional risk information can be used to guide the destination where the user departs. If the user is using the travel recommendation service of this system for the first time, the system of the present invention will provide the user with a virtual risk score representing the possible risks and initialize it as the risk score of the user's location in the regional risk information. The user can also manually modify it according to their own situation, and this risk score will only be used locally on the user terminal in the subsequent process.

[0092] S2: The user terminal obtains a personalized travel mode prediction service that meets the user-level preferences through federated learning, including:

[0093] S2.1: The user terminal pulls the current first global model to the local from the server. The first global model has certain prediction performance and can be locally trained while the user uses the travel service, and uploads the second global model to the server for aggregation after training. The first global model after federated learning can be used as the second global model.

[0094] S2.2: After the server accepts the second global model and performs federated learning aggregation, it tests the second global model on the public dataset. The second global model that passes the test is used as the third global model, and the third global model is returned to the user terminal to provide the travel mode recommendation service for the user, and the optimization process based on federated learning ends. Subsequently, the third global model can be directly used for prediction. If the second global model fails the test, the second global model is returned to the user terminal, and the user terminal is informed that the second global model still needs to be trained, and then waits for the user terminal to upload the second global model again for the next aggregation task.

[0095] S2.3: The user uses the third global model through the user terminal to predict the travel mode. If it does not meet the user's travel preferences, the user can make a manual selection, and the third global model enters the next round of training; if it meets, it enters S3.

[0096] S3: The user terminal obtains the best travel route considering risk factors using the regional risk information and the travel mode, including:

[0097] S3.1: The user terminal uses the third global model to predict the personalized travel mode, provides the starting point and destination of the travel to the server, in order to obtain relevant data from the server.

[0098] S3.2: The server returns candidate road network information according to the starting point and destination of the user's travel for the user to plan locally on the user terminal. If the user uses the travel recommendation service for the first time, the server can also return the path planning algorithm for area avoidance to the user terminal.

[0099] S3.3: The user terminal uses the local regional risk information and candidate road network information, and uses the path planning algorithm for regional avoidance to plan a safe route.

[0100] S4: After the user travels, the user terminal will extract part of the travel trajectory according to the travel category, perform fuzzification, and finally upload it to the server to update the regional risk information. The specific process of fuzzifying the travel trajectory can refer to the above embodiments.

[0101] Refer to Figure 5 , the embodiment of the present invention provides a secure personalized travel recommendation method based on federated learning and regional avoidance, including:

[0102] Step S200: Obtain regional risk information from the server through the user terminal, and obtain the destination determined by the user according to the regional risk information;

[0103] Step S210: Obtain the first global model from the server through the user terminal, perform federated learning on the first global model according to the regional risk information and the destination, and upload the first global model after federated learning training to the server. The first global model after federated learning training is used as the second global model;

[0104] Step S220: The server performs federated learning aggregation on the second global model, tests the second global model after federated learning aggregation on the public dataset, and sends the second global model that meets the preset test requirements and the path planning algorithm to the user terminal. The second global model that meets the preset test requirements is used as the third global model;

[0105] Step S230: The user terminal uses the third global model and combines the regional risk information to predict the user's travel mode, and uses the path planning algorithm and combines the travel mode to predict the first travel route of the user from the specified starting point to the destination.

[0106] The specific implementation process of the method of the present invention can refer to the operation process of the system of the present invention described above, and will not be repeated here.

[0107] The embodiment of the present invention also discloses a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes Figure 5 the method shown.

[0108] In some alternative embodiments, the functions / operations recited in the block diagrams may not occur in the order presented in the operational illustrations. For example, depending on the functions / operations involved, two blocks shown in succession may actually be executed substantially concurrently or the blocks may sometimes be executed in reverse order. Further, the embodiments presented and described in the flowcharts of the present invention are provided by way of example in order to provide a more thorough understanding of the technology. The disclosed methods are not limited to the operations and logical flows presented herein. Alternative embodiments are envisioned in which the order of various operations is altered and in which sub-operations described as part of a larger operation are performed independently.

[0109] In addition, although the present invention has been described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the described functions and / or features may be integrated in a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It should also be understood that a detailed discussion of the actual implementation of each module is not necessary for an understanding of the present invention. Rather, given the attributes, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the modules will be understood within the ordinary skill of an engineer. Thus, those of ordinary skill in the art will be able to implement the present invention as set forth in the claims without undue experimentation. It should also be understood that the particular concepts disclosed are illustrative only and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0110] If the functions are implemented in the form of software functional units and sold or used as independent products, they may be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present invention, in essence or the part that contributes to the prior art or part of the technical solution, may be embodied in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes, such as a USB flash drive, a portable hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc.

[0111] The logic and / or steps represented in the flowchart or otherwise described herein can, for example, be considered as a definable sequence of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device. As used in this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with the instruction execution system, apparatus, or device.

[0112] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection portion having one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, as the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then storing it in a computer memory.

[0113] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0114] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0115] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the claims and their equivalents.

[0116] The above has specifically described the preferred embodiments of the present invention, but the present invention is not limited to the described embodiments. Those skilled in the art can also make various equivalent deformations or substitutions without departing from the spirit of the present invention, and these equivalent deformations or substitutions are all included within the scope defined by the claims of the present invention.

Claims

1. A secure personalized travel recommendation system based on federated learning and regional avoidance, characterized in that, It includes a user terminal and a server; The user terminal is used for: Obtaining regional risk information from the server, and obtaining a destination determined by the user according to the regional risk information; Obtaining a first global model from the server, performing federated learning on the first global model according to the regional risk information and the destination, and uploading the first global model after federated learning training to the server, and using the first global model after federated learning training as a second global model; Predicting the user's travel mode by using a third global model and combining the regional risk information, and predicting a first travel route from a specified starting point to the destination by using a path planning algorithm and combining the travel mode; The server is used for: Performing federated learning aggregation on the second global model, testing the second global model after federated learning aggregation on a public dataset, and sending the second global model that meets the preset test requirements and the path planning algorithm to the user terminal, and using the second global model that meets the preset test requirements as a third global model; The user terminal is further used for uploading the starting point, the destination and the first travel route to the server for the server to update the regional risk information according to the starting point, the destination and the first travel route; The user terminal includes: A travel information expansion unit for expanding at least one of the starting point, the destination or the first travel route outward by a set range; A travel information uploading unit for uploading at least one of the starting point, the destination or the first travel route after the expansion range to the server; The user terminal includes: A travel route prediction unit for predicting a first appearance route of the user from the starting point to the destination by using the path planning algorithm according to a preset path calculation formula and combining the travel mode; The path calculation formula is: Cost road = Length road + Abs(Risk user - Risk road ) × α × β Among them, Cost road represents the cost of the candidate path, Length road represents the length of the candidate path, Abs() represents the absolute value function, Risk user represents the virtual risk score of the user, Risk road represents the risk score of the candidate path determined according to the regional risk information, α represents the risk concern coefficient that can be set by the user, and β represents whether the user uses a preset target travel mode.

2. The secure personalized travel recommendation system based on federated learning and regional avoidance according to claim 1, wherein The server is further used for determining candidate road network information according to the starting point and the destination, and sending the candidate road network information to the user terminal for the user to determine a second travel route according to the candidate road network information.

3. The secure personalized travel recommendation system based on federated learning and regional avoidance according to claim 1, wherein The server is further used for sending the second global model that does not meet the preset test requirements to the user terminal for the user terminal to perform federated learning on the second global model that does not meet the preset test requirements.

4. A secure personalized travel recommendation method based on federated learning and regional avoidance, characterized in that, It includes: Obtaining regional risk information from the server through the user terminal, and obtaining a destination determined by the user according to the regional risk information; Obtaining a first global model from the server through the user terminal, performing federated learning on the first global model according to the regional risk information and the destination, and uploading the first global model after federated learning training to the server, and using the first global model after federated learning training as a second global model; The second global model is aggregated through federated learning on the server side, and the second global model after federated learning aggregation is tested on a public dataset. The second global model that meets the preset test requirements and the path planning algorithm are sent to the user terminal, and the second global model that meets the preset test requirements is used as the third global model; The user terminal uses the third global model and combines the regional risk information to predict the travel mode of the user, and uses the path planning algorithm and combines the travel mode to predict the first travel route of the user from a specified starting point to the destination; The method further includes: The user terminal uploads the starting point, the destination, and the first travel route to the server side, so that the server side can update the regional risk information according to the starting point, the destination, and the first travel route; The uploading of the starting point, the destination, and the first travel route to the server side by the user terminal includes: The user terminal expands at least one of the starting point, the destination, or the first travel route by a set range; The user terminal uploads at least one of the starting point, the destination, or the first travel route after the expansion range to the server side; The predicting of the first travel route of the user from a specified starting point to the destination by using the path planning algorithm and combining the travel mode includes: The user terminal uses the path planning algorithm according to a preset path calculation formula and combines the travel mode to predict the travel route of the user from the starting point to the destination; The path calculation formula is: Cost road = Length road + Abs(Risk user - Risk road ) × α × β Among them, Cost road represents the cost of the candidate path, Length road represents the length of the candidate path, Abs() represents the absolute value function, Risk user represents the virtual risk score of the user, Risk road represents the risk score of the candidate path determined according to the regional risk information, α represents the risk concern coefficient that can be set by the user, and β represents whether the user uses a preset target travel mode.

Images