Method, device, equipment and storage medium for confirming rationality of driving route

Abstract

The present disclosure relates to a method, device, equipment and storage medium for confirming the rationality of a driving route. The method comprises: obtaining actual travel information from a starting point to a destination; estimating the navigation route and travel time from the starting point to the destination at a preset time based on the actual travel information; when the actual travel information is complained, at least the navigation route and travel time are used as input features of a deep learning framework to determine the rationality judgment result of the actual travel. The method for confirming the rationality of a driving route provided by the present disclosure can improve the accuracy of confirming the rationality of a driving route.

Description

Technical Field

[0001] The present disclosure relates to the field of information technology, and in particular to a method, device, equipment and storage medium for confirming the rationality of a driving route. Background Art

[0002] With the continuous development of science and technology, smart terminals have become an indispensable device in people's daily life. Users can install various types of applications (Application, APP) on smart terminals to meet different needs of users.

[0003] Travel applications, such as online car-hailing applications, are widely used. Usually, online car-hailing applications can plan a navigation route from the starting point to the destination, and the driver can provide online car-hailing services to users along the navigation route or based on experience. In actual situations, passengers often think that the driver's driving route is unreasonable, resulting in the fare not meeting the passenger's expectations, which leads to passenger complaints. Therefore, online car-hailing platforms need to have the technical ability to determine whether the driver's driving route is unreasonable. Summary of the invention

[0004] In order to solve the above technical problems or at least partially solve the above technical problems, the present disclosure provides a method, device, equipment and storage medium for confirming the rationality of a driving route, which can improve the accuracy of confirming the rationality of a driving route.

[0005] In a first aspect, an embodiment of the present disclosure provides a method for confirming the rationality of a driving route, the method comprising:

[0006] Get the actual travel information from the starting point to the destination;

[0007] Based on the actual travel information, estimate the navigation route and travel time from the starting point to the destination at the preset time;

[0008] When complaints are made about actual travel information, at least the navigation route and travel time are used as input features of the deep learning framework to determine the rationality of the actual travel.

[0009] In a second aspect, an embodiment of the present disclosure provides a device for confirming the rationality of a driving route, the device comprising:

[0010] The acquisition module is used to obtain the actual travel information from the starting point to the destination;

[0011] An estimation module is used to estimate the navigation route and travel time from the starting point to the destination at a preset time based on the actual travel information;

[0012] The determination module is used to determine the rationality judgment result of the actual trip by using at least the navigation route and travel time as input features of the deep learning framework when the actual trip information is complained.

[0013] In a third aspect, an embodiment of the present disclosure provides an electronic device, including:

[0014] Memory;

[0015] Processor; and

[0016] Computer programs;

[0017] The computer program is stored in the memory and is configured to be executed by the processor to implement the method as described in the first aspect.

[0018] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the method described in the first aspect.

[0019] The method, device, equipment and storage medium for confirming the rationality of a driving route provided by the embodiments of the present disclosure obtain actual travel information from a starting point to a destination; based on the actual travel information, estimate the navigation route and travel time from the starting point to the destination at a preset time; when a complaint is made about the actual travel information, at least the navigation route and travel time are used as input features of a deep learning framework to determine the rationality judgment result of the actual travel; in other words, at least the estimated navigation route and travel time from the starting point to the destination are used as input features of the deep learning framework, and the rationality judgment result of the actual travel is determined through the deep learning framework, thereby achieving the purpose of determining the rationality judgment result of the actual travel based on technical means and improving the accuracy of determining the rationality judgment result of the actual travel. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0022] Figure 1 A flow chart of a method for confirming the rationality of a driving route provided in an embodiment of the present disclosure;

[0023] Figure 2A flow chart of a method for confirming the rationality of a driving route provided in an embodiment of the present disclosure;

[0024] Figure 3 A flow chart of a method for confirming the rationality of a driving route provided in an embodiment of the present disclosure;

[0025] Figure 4 A schematic diagram of a deep learning framework provided in an embodiment of the present disclosure;

[0026] Figure 5 A flow chart of a method for confirming the rationality of a driving route provided by another embodiment of the present disclosure;

[0027] Figure 6 A flowchart diagram provided for another embodiment of the present disclosure;

[0028] Figure 7 A schematic diagram of the structure of a device for confirming the rationality of a driving route provided in an embodiment of the present disclosure;

[0029] Figure 8 A schematic diagram of the structure of an electronic device embodiment provided by the present disclosure. DETAILED DESCRIPTION

[0030] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0031] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0032] Normally, online car-hailing applications can plan a navigation route from the starting point to the destination, and the driver can provide online car-hailing services to users along the navigation route or based on experience. However, in actual situations, passengers often think that the driver's driving route is unreasonable, for example, the driver deliberately takes a detour, resulting in the cost not meeting the passenger's expectations, thus causing complaints from passengers. When determining whether the driver has taken a detour, that is, when determining whether the driver's formal route is reasonable, the conventional practice is to determine it based on static information such as the passenger's historical evaluation of the driver, estimated mileage, actual mileage, estimated time, actual time, and travel expenses. Since the granularity of the information referenced by the conventional practice is not fine enough and the dimension is single, and the passenger's evaluation of the driver is somewhat subjective, the accuracy of the determination result obtained is not high.

[0033] In view of the above problems, the embodiments of the present disclosure provide a method for confirming the rationality of a driving route, so as to objectively determine whether the driver has a detour problem by means of technical means. The method is introduced below in conjunction with a specific embodiment.

[0034] Figure 1 This is a flow chart of a method for confirming the rationality of a driving route provided in an embodiment of the present disclosure. This embodiment is applicable to situations where the rationality of a driving route is to be determined. The method can be executed by a device for confirming the rationality of a driving route. The device can be implemented in software and / or hardware. The device can be configured in an electronic device, such as a server. Figure 1 As shown, the specific steps of this method are as follows:

[0035] S101. Acquire actual travel information from a starting point to a destination.

[0036] Among them, the actual trip information may include at least one of the order generation time corresponding to the actual trip, the actual departure time corresponding to the actual trip, the actual driving route, and the actual time. For example, a passenger calls a car through an online car-hailing application at 10:55 a.m. on a certain day. Assuming that the system delay is ignored, a car-hailing order is generated at 10:55. At 11 o'clock, the online car-hailing picks up the passenger at place A, that is, the passenger gets on the car and starts the journey. At 11:30, the online car-hailing delivers the passenger to the destination place B. Assuming that there are three routes from place A to place B, namely route 1, route 2 and route 3, in this trip, the online car-hailing carries the passenger from place A to place B via route 1, that is, the actual driving route is route 1. In the above application scenario, the actual trip information may include: order generation time-10:55, actual departure time-11 o'clock, actual driving route-route 1, actual time-half an hour. That is, the actual driving route refers to the route taken during the trip, which may be the optimal route recommended by the software, the route navigated by the driver, or the route specified by the passenger.

[0037] S102: based on the actual travel information, estimating the navigation route and travel time from the starting point to the destination at a preset time.

[0038] The preset time may be one or more times near the order generation time or the actual departure time included in the actual trip information. For example, if the actual departure time is 11:00, the preset time may be 10:50 and / or 11:10; or, the preset time may be 10:50, 10:40, 11:10, and 11:20, etc. The number of preset times can be set.

[0039] By estimating the navigation route and travel time from the starting point to the destination at one or more times near the order generation time or the actual departure time, multiple reference results can be obtained, and the multiple reference results are used to assist in determining whether the driving route in the actual trip information is reasonable. For example, multiple reference results can be used to determine whether the actual time is relatively abrupt. By obtaining more realistic conditions in the actual trip based on these dynamically changing status information (such as navigation routes and travel times vary with road conditions, such as traffic accidents, traffic control, snowy road closures, etc., which will cause different navigation routes and travel times planned by the algorithm), it is possible to objectively and truly reflect whether the driver's driving route is reasonable.

[0040] Continuing with the above application scenario as an example, for example, it is estimated that at 10:50, it takes 10 minutes to drive from place A to place B via Route 2, and at 11:10, it takes 8 minutes to drive from place A to place B via Route 2. However, the actual driving route is Route 1, and the actual driving time is half an hour. By referring to the estimated results, it can be determined that the driver is more likely to take a detour and the driving route is more likely to be unreasonable.

[0041] S103. When a complaint is made about the actual travel information, at least the navigation route and the travel time are used as input features of the deep learning framework to determine the rationality judgment result of the actual travel.

[0042] Among them, the rationality judgment results of the actual trip may include information such as whether the driver is responsible / not responsible, whether the passenger is responsible / not responsible, or whether the service provider is responsible / not responsible.

[0043] It is understandable that the deep learning framework is a trained deep learning framework. In order to deeply explore the relationship between the navigation route and travel time obtained by estimation and the actual driving route, the embodiment of the present disclosure uses the deep learning framework, at least the navigation route and travel time as the input features of the deep learning framework, to determine the rationality judgment result of the actual trip, which can improve the objectivity and accuracy of the rationality judgment result of the actual trip. At the same time, by estimating the dynamic information that changes with time and road conditions (such as the navigation route and travel time vary with different road conditions, such as traffic accidents, traffic control, snowy road closures and other factors will make the navigation route planned by the algorithm different and the travel time different) as the input features of the deep learning framework, it can directly and truly reflect the various conditions in the actual trip, and can more objectively and truly reflect the actual situation of the driver's detour, thereby improving the accuracy of the confirmation of the rationality of the driving route. In particular, the driver's detour refers to the process in which the actual walking trajectory (actual walking journey) that does not conform to the planned route actually occurs, resulting in the phenomenon that the cost does not meet expectations or the passengers are dissatisfied.

[0044] In the actual taxi-hailing process, at the end of the trip, the passenger may complain about the trip because the actual trip took a long time or the fare was high, such as calling a complaint hotline or making an online complaint through an online taxi-hailing application. When a complaint is received from a passenger, the actual trip information of the complained-about trip is obtained, and based on the actual trip information, the navigation route and travel time from the starting point to the destination at the preset time are estimated. At least the navigation route and travel time are used as input features of the deep learning framework to determine the rationality judgment result of the actual trip.

[0045] The disclosed embodiment obtains actual travel information from a starting point to a destination; estimates a navigation route and travel time from the starting point to the destination at a preset time based on the actual travel information; when a complaint is made about the actual travel information, at least the navigation route and travel time are used as input features of a deep learning framework to determine a result of a rationality judgment of the actual travel; in other words, at least the estimated navigation route and travel time from the starting point to the destination are used as input features of a deep learning framework, and a result of a rationality judgment of the actual travel is determined through the deep learning framework, thereby achieving the purpose of determining a result of a rationality judgment of the actual travel based on technical means and improving the accuracy of determining a result of a rationality judgment of the actual travel.

[0046] Figure 2 A flowchart of a method for confirming the rationality of a driving route provided by another embodiment of the present disclosure. In this embodiment, an optional implementation is provided for the above step S102 "estimate the navigation route and travel time from the starting point to the destination at a preset time based on the actual travel information". Figure 2 As shown, the specific steps of this method are as follows:

[0047] S201. Obtain actual travel information from a starting point to a destination.

[0048] The actual itinerary information includes at least one of the following: an order generation time corresponding to the actual itinerary and an actual departure time corresponding to the actual itinerary.

[0049] S202: Determine one or more preset times before the order generation time or the actual departure time; and / or determine one or more preset times after the order generation time or the actual departure time.

[0050] In one embodiment, one or more preset times before the order generation time or the actual departure time are determined. For example, if the actual departure time is 11:00 a.m., the preset time may be 10:50, 10:40, 10:30, etc.

[0051] In another embodiment, one or more preset times after the order generation time or the actual departure time are determined. For example, if the actual departure time is 11:00 a.m., the preset time may be 11:10, 11:20, 11:30, etc.

[0052] In another embodiment, one or more preset times before the order generation time or the actual departure time are determined, and one or more preset times after the order generation time or the actual departure time are determined. For example, if the actual departure time is 11:00 a.m., the preset times may be 10:50, 10:40, 10:30, 11:10, 11:20, 11:30, etc.

[0053] By estimating the navigation route and travel time from the starting point to the destination at one or more times near the order generation time or the actual departure time, multiple reference results can be obtained, and the multiple reference results are used to assist in determining whether the navigation route in the actual trip information is reasonable. Continuing with the above application scenario as an example, for example, it is estimated that at 10:50, it takes 10 minutes to drive from place A to place B via Route 2, and at 11:10, it takes 8 minutes to drive from place A to place B via Route 2, while the actual driving route is Route 1, and the actual travel time is half an hour. Then, by referring to the estimated results, it can be determined that the driver is more likely to take a detour.

[0054] S203: for each of the more than one preset times, obtain a navigation route and travel time from the starting point to the destination at the preset time through future travel prediction.

[0055] Among them, future trip prediction refers to: estimating the navigation route and travel time at a future time (from the starting point to the destination), and predicting possible congested sections and dynamic event sections (such as traffic control) on the route, as well as the restricted areas that are effective at that time.

[0056] For example, if the actual departure time is 11:00 a.m., and the preset times are 10:50, 10:40, 10:30, 11:10, 11:20, and 11:30 a.m., then the estimated navigation route and travel time from place A to place B at 10:50 (recorded as navigation route 1 and travel time 1); the estimated navigation route and travel time from place A to place B at 10:40 (recorded as navigation route 2 and travel time 2); the estimated navigation route and travel time from place A to place B at 10:30 (recorded as navigation route 2 and travel time 2); The navigation route and travel time from place A to place B (recorded as navigation route 3 and travel time 3); at 11:10, the navigation route and travel time from place A to place B (recorded as navigation route 4 and travel time 4); at 11:20, the navigation route and travel time from place A to place B (recorded as navigation route 5 and travel time 5); at 11:30, the navigation route and travel time from place A to place B (recorded as navigation route 6 and travel time 6).

[0057] S204: When a complaint is made about the actual travel information, at least the navigation route and the travel time are used as input features of the deep learning framework to determine the rationality judgment result of the actual travel.

[0058] Exemplarily, at least the above-mentioned navigation route 1 and travel time 1, navigation route 2 and travel time 2, navigation route 3 and travel time 3, navigation route 4 and travel time 4, navigation route 5 and travel time 5, and navigation route 6 and travel time 6 are used as input features of the deep learning framework to determine the rationality judgment result of the actual itinerary.

[0059] Specifically, the implementation methods and specific principles of S201 and S101, S204 and S103 are consistent, and will not be repeated here.

[0060] This embodiment provides an optional implementation method for the above step S102 "estimating the navigation route and travel time from the starting point to the destination at the preset time based on the actual travel information", which specifically determines one or more preset times before the order generation time or the actual departure time, and / or determines one or more preset times after the order generation time or the actual departure time, and for each of the more than one preset times, obtains the navigation route and travel time from the starting point to the destination at the preset time through future travel estimation. By estimating the navigation route and travel time from the starting point to the destination at one or more times near the order generation time or the actual departure time, multiple reference results can be obtained. The multiple reference results are used to assist in determining whether the navigation route in the actual travel information is reasonable. Since the multiple reference results can more objectively reflect the various conditions of the actual road conditions, the combination of the multiple reference results helps to improve the accuracy of confirming the rationality of the actual driving route.

[0061] Figure 3A flowchart of a method for confirming the rationality of a driving route provided by another embodiment of the present disclosure. In this embodiment, an optional implementation is provided for the above step S103 "when the actual travel information is complained, at least the navigation route and travel time are used as input features of the deep learning framework to determine the rationality judgment result of the actual travel". Figure 3 As shown, the specific steps of this method are as follows:

[0062] S301, obtaining actual travel information from the starting point to the destination.

[0063] S302: based on the actual travel information, estimate the navigation route and travel time from the starting point to the destination at the preset time.

[0064] Specifically, the implementation methods and specific principles of S301 and S101, S302 and S202-S203 are consistent, and will not be repeated here.

[0065] S303: Obtain multi-dimensional information according to the actual driving route and the navigation route.

[0066] The multi-dimensional information includes at least one of the following:

[0067] The route length of the navigation route, the difference in length between the navigation route and the actual driving route, the number of overlapping road units in the navigation route and the actual driving route, the sum of the lengths of the overlapping road units, the number of non-overlapping road units in the navigation route and the actual driving route, and the sum of the lengths of the non-overlapping road units. Among them, the road unit refers to a directed logical road unit divided according to the actual road, corresponding to a section of road in the road network map, which can be recorded as a link. The multi-dimensional information can also include the number of non-overlapping road units in the actual driving route and the navigation route, as well as the sum of the lengths of the non-overlapping road units.

[0068] Here are some examples to illustrate the above multi-dimensional information:

[0069] Assume that the actual departure time of the passenger and the driver from the starting point (e.g., place A) is 11 o'clock, and the actual time consumed to reach the destination (e.g., place B) is 1 hour. Select 6 time points within the hour before 11 o'clock, for example, select a time point every 10 minutes, and the 6 time points are 10:00, 10:10, 10:20, 10:30, 10:40, and 10:50. Similarly, select 6 time points within the hour after 11 o'clock, which are 11:10, 11:20, 11:30, 11:40, 11:50, and 12:00. Since the server can store the traffic information of the entire road network at each moment, the server can plan the future itinerary based on the traffic information of the entire road network at 10:00 and 10:00, thereby estimating the navigation route of the passengers and drivers from place A to place B at 10:00 (the navigation route can be considered to be the optimal route, and the navigation route is recorded as navigation route 1) and the travel time (the travel time is recorded as travel time 1). The server can also plan the future itinerary based on the traffic information of the entire road network at 10:10 and 10:10, and estimate the navigation route of the passengers and drivers from place A to place B at 10:10 (the navigation route is recorded as navigation route 2) and the travel time (the travel time is recorded as travel time 2). By analogy, 12 navigation routes and 12 travel times are finally obtained. It can be understood that the 12 navigation routes may be different, and the 12 travel times may also be different.

[0070] In the above 12 navigation routes, different navigation routes may include different road units (link will be used to represent road units later) and the total number of Links. For example, the above navigation route 1 includes 3 Links, namely LinkA, LinkB and LinkC, and the above navigation route 2 includes 4 Links, namely LinkB, LinkC, LinkD and LinkE. It is understandable that the congestion and time consumption corresponding to the same Link in different navigation routes may also be different, because the real-time road conditions are changing at any time. For example, LinkB may not be congested in navigation route 1, and the time consumption through LinkB is relatively short; but LinkB may become congested in navigation route 2, and the time consumption through LinkB is relatively long. The above objective factors enable the disclosed scheme to obtain various more realistic conditions in the actual journey based on these dynamically changing status information, and can more objectively reflect whether the driving route taken by the driver is reasonable.

[0071] Based on the above-mentioned application scenarios, the number of overlapping road units in the above-mentioned navigation route and the actual driving route refers to: for example, navigation route 1 includes LinkA, LinkB and LinkC, and the actual driving route includes LinkB, LinkC and LinkD, then the overlapping road units in the navigation route and the actual driving route include LinkB and LinkC, and their number is 2. The sum of the lengths of the overlapping road units refers to the sum of the road lengths of LinkB and LinkC. The road unit that does not overlap between the navigation route and the actual driving route is LinkA, and its number is 1. The sum of the lengths of the non-overlapping road units refers to the road length of LinkA. The route length of the navigation route refers to the sum of the road lengths of LinkA, LinkB and LinkC. The length difference between the navigation route and the actual driving route refers to the difference between the sum of the road lengths of LinkA, LinkB and LinkC and the sum of the road lengths of LinkB, LinkC and LinkD. The road unit that does not overlap between the actual driving route and the navigation route is LinkD, and its number is 1, and the sum of the lengths of the non-overlapping road units is the road length of LinkD.

[0072] S304: Use the multi-dimensional information and travel time as input features of the bidirectional long short-term memory network to obtain output features of the bidirectional long short-term memory network.

[0073] S305. Use at least one of the order information and user portrait information corresponding to the actual itinerary and the output features of the bidirectional long short-term memory network as input features of the multi-layer perceptron to determine the rationality judgment result of the actual itinerary.

[0074] The rationality judgment result of the actual trip may include the probability that the driver is responsible and the probability that the driver is not responsible. If the probability that the driver is responsible is greater than the probability that the driver is not responsible, then the driver is determined to be responsible, that is, the driver has taken a detour. If the probability that the driver is responsible is less than the probability that the driver is not responsible, then the driver is determined to be not responsible, that is, the driver did not take a detour.

[0075] For example, see Figure 4 A schematic diagram of a deep learning framework is shown, wherein the deep learning framework includes a bidirectional long short-term memory network (BIdirectionalLong Short-term Memory, BILSTM) 430 and a multi-layer perceptron (Multi-layer Perceptron, MLP) 440.

[0076] Multi-dimensional information such as Figure 4The matrix 410 and the travel time 420 in are used as input features of the bidirectional long short-term memory network 430 to obtain the output features of the bidirectional long short-term memory network 430. At least one of the order information and the user profile information corresponding to the actual trip and the output features of the bidirectional long short-term memory network are used as input features of the multilayer perceptron 440, and the output of the multilayer perceptron 440 is the rationality judgment result of the actual trip.

[0077] like Figure 4 As shown, the bidirectional long short-term memory network 430 includes an attention mechanism layer Attention.

[0078] In some embodiments, the matrix 410 of multi-dimensional information may be a 13*6 matrix, i.e., the number of matrix rows m=13, and the number of columns n=6. The above application scenario is continued to be cited, and the physical meaning of each dimension is explained by taking the first row b11, b12...b1n, which is a total of 6 dimensions, as an example. b11 represents the route length of navigation route 1 of passengers and drivers from A to destination B at the preset time 10:00 estimated by future itinerary planning. b12 represents the difference between the route length of navigation route 1 and the actual driving route length. b13 represents the number of links that overlap between navigation route 1 and the actual driving route. For example, navigation route 1 includes 30 links, and the actual driving route includes 20 links. There are 5 links that overlap between the 30 links and the 20 links. b14 represents the sum of the lengths of the links that overlap between navigation route 1 and the actual driving route, such as the sum of the lengths of the 5 links mentioned above. b15 represents the sum of the lengths of the 25 non-overlapping links obtained by subtracting the 5 overlapping links from the 30 links included in navigation route 1. b16 indicates that the number of non-overlapping links obtained by subtracting the 5 overlapping links from the 30 links included in navigation route 1 is 25. Similarly, the second row b21, b22, ... b2n of the matrix 410 of multi-dimensional information is 6 dimensions obtained based on the actual driving route and the estimated navigation route 2 of the passengers and drivers when they depart at the preset time of 10:10. The 6 dimensions are similar to the 6 dimensions of the first row b11, b12 ... b1n mentioned above. It can be understood that for the actual departure time of 11:00, there is an actual driving route; in addition, a navigation route and travel time can be estimated based on the road condition information at 11:00. The navigation route and the actual driving route may be different, and the travel time may be different from the actual time. Therefore, the 7th row of the matrix 410 of multi-dimensional information is 6 dimensions obtained based on the navigation route and the actual driving route at the preset time of 11:00. In addition, the 8th to 13th rows of the matrix 410 are 6 rows in total, which are obtained based on the navigation routes and actual driving routes corresponding to 6 time points within 1 hour after 11:00, which is similar to the above process and will not be further explained.

[0079] In other embodiments, the matrix 410 may also be a 13*8 matrix, i.e., a matrix of 13 rows and 8 columns, and the elements of the first row may include b17 and b18 in addition to the aforementioned b11...b16. Among them, b17 represents the sum of the route lengths of the 15 non-overlapping Links obtained by subtracting the 5 overlapping Links from the 20 Links included in the actual driving route. b18 represents the number of non-overlapping Links obtained by subtracting the 5 overlapping Links from the 20 Links included in the actual driving route, i.e., 15. Similarly, the physical meanings of the two dimensions added in other rows can refer to b17 and b18, and will not be repeated here.

[0080] In addition, if Figure 4 As shown, the travel time t1 in the input feature of the bidirectional long short-term memory network represents the difference between the actual travel time and the estimated travel time at the time point 10:00 (may be a positive value or a negative value), t2 represents the difference between the actual travel time and the estimated travel time at the time point 10:10,…, t7 represents the difference between the actual travel time and the estimated travel time at the time point 11:00 (may be 0 or not 0), t8 represents the difference between the actual travel time and the estimated travel time at the time point 11:10, and so on, t13 represents the difference between the actual travel time and the estimated travel time at the time point 12:00.

[0081] Furthermore, t1…t13 and the above matrix 410 can be concatenated to obtain a new matrix (the new matrix is ​​denoted as matrix B). The specific concatenation method can be to use t1…t13 as the first column of matrix B, the first column of matrix 410 as the second column of matrix B, the second column of matrix 410 as the third column of matrix B, the third column of matrix 410 as the fourth column of matrix B, and so on, and the sixth column of matrix 410 as the seventh column of matrix B. In other words, matrix B is a matrix with 13 rows and 7 columns. Further, the first row of matrix B is denoted as S1, the second row of matrix B is denoted as S2, and so on, and the 13th row of matrix B is denoted as S13. S1…S13 are used as input features of the bidirectional long short-term memory network. The output features of the bidirectional long short-term memory network include the relationship between the actual time and the estimated 12 travel times, the relationship between the actual driving route and the estimated 12 navigation routes, the relationship between the actual time and the actual driving route, and the relationship between the estimated 12 travel times and the estimated 12 navigation routes. By referring to the above dynamically changing state information, various more realistic conditions in the actual journey can be obtained, which can more objectively reflect whether the driving route taken by the driver is reasonable.

[0082] Furthermore, in addition to the output features of the bidirectional long short-term memory network 430, the input features of the multilayer perceptron MLP 440 also include at least one of the order information and user profile information corresponding to the actual trip. The order information corresponding to the actual trip may include: at least one of the service provider number (such as the platform provider number of the online car-hailing service), estimated mileage, actual mileage, estimated price, actual price, estimated time and actual time; the estimated mileage, estimated price and estimated time may be provided by a third party, or may be obtained through future trip estimates. User portrait information may specifically be user portrait information of drivers and / or passengers, including, for example: driver's historical order statistics (such as the total number of driver orders), passenger's historical order statistics (such as the total number of passenger orders), driver's complaint information statistics, passenger's complaint information statistics, driver's number of orders in the past week, accountability information statistics (the rationality judgment result or disposal plan of the actual trip given in response to the passenger's historical complaints, which can be pushed to the passenger's online car-hailing client), driver reporting information (such as the driver's pre-reported passenger's drinking) and complaint statistics (such as the driver being complained about, the platform compensating the passenger, and the driver's complaint behavior that he thinks is unreasonable). By adding user portrait information, the reference information can be enriched, thereby further improving the accuracy of the rationality judgment result of the actual trip.

[0083] Understandably, Figure 4 The deep learning framework shown can be understood as a fault-determination system. The deep learning framework needs to be trained in advance before use. In the training phase of the deep learning framework, multiple sample data are required, and each sample data can be a historical travel record of a driver who has been manually marked or manually determined to have a detour behavior. The historical travel record includes the actual departure time, the actual driving route, the actual time, etc. Further, according to the historical travel record and the future travel plan, a first matrix similar to the above-mentioned matrix 410 is obtained. Further, t1…t13 is calculated, and according to t1…t13 and the first matrix, a second matrix similar to the above-mentioned matrix B is obtained. Each row of the second matrix is ​​processed by a bidirectional long short-term memory network to obtain the output features of the bidirectional long short-term memory network. The output features of the bidirectional long short-term memory network, the order information corresponding to the historical travel record, and the user portrait information are used as the input of the multi-layer perceptron MLP, and the multi-layer perceptron MLP outputs a classification label (i.e., the probability that the driver is responsible and / or the probability that the driver is not responsible). It is understandable that the classification labels obtained in the training stage may be inaccurate. Furthermore, the deep learning framework is trained based on the classification labels output by the multi-layer perceptron MLP and the correct results (for example, the driver is responsible because the sample data has been manually labeled or manually determined as having a history of the driver taking detours).

[0084] This embodiment provides an optional implementation method for the above step S103 "When the actual travel information is complained, at least the navigation route and travel time are used as input features of the deep learning framework to determine the rationality judgment result of the actual travel", which is: multi-dimensional information is obtained according to the actual driving route and the navigation route; the multi-dimensional information and travel time are used as input features of the bidirectional long short-term memory network to obtain the output features of the bidirectional long short-term memory network; at least one of the order information and user profile information corresponding to the actual travel and the output features of the bidirectional long short-term memory network are used as input features of the multilayer perceptron to determine the rationality judgment result of the actual travel. By adding user profile information, the reference information can be enriched, thereby further improving the accuracy of the rationality judgment result of the actual travel.

[0085] Figure 5 A flow chart of a method for confirming the rationality of a driving route provided in another embodiment of the present disclosure. In this embodiment, after determining the rationality judgment result of the actual trip, the following steps are added: generating a treatment plan based on the rationality judgment result of the actual trip; pushing the treatment plan to the passenger terminal; upon receiving confirmation information of the treatment plan returned by the passenger terminal, generating sample data, the sample data including the actual trip information; and training the deep learning framework based on the sample data. By continuously training the deep learning framework with actual existing data as new sample data, the performance of the deep learning framework can be continuously improved, so that it can give increasingly accurate rationality judgment results of the actual trip. Figure 5 As shown, the specific steps of this method are as follows:

[0086] S501. Obtain actual travel information from the starting point to the destination.

[0087] S502: based on the actual travel information, estimate the navigation route and travel time from the starting point to the destination at the preset time.

[0088] Specifically, the implementation methods and specific principles of S501 and S101, S502 and S202-S203 are consistent, and will not be repeated here.

[0089] S503. Obtain multi-dimensional information according to the actual driving route and the navigation route, use the multi-dimensional information and the travel time as input features of a bidirectional long short-term memory network, and obtain output features of the bidirectional long short-term memory network.

[0090] S504. Use at least one of the order information and user portrait information corresponding to the actual itinerary and the output features of the bidirectional long short-term memory network as input features of the multi-layer perceptron to determine the rationality judgment result of the actual itinerary.

[0091] S505. Generate a treatment plan based on the rationality judgment result of the actual journey, push the treatment plan to the passenger terminal, and upon receiving confirmation information of the treatment plan returned by the passenger terminal, generate sample data, the sample data including the actual journey information.

[0092] Among them, the handling plan is, for example, if the rationality judgment result of the actual journey shows that the driver is responsible, then the passenger will be given certain compensation.

[0093] S506. Train the deep learning framework according to the sample data.

[0094] By continuously acquiring sample data to continuously train the deep learning framework, the performance of the deep learning framework can be continuously improved, making the rationality judgment results of the actual itinerary output by it more and more accurate.

[0095] In a specific application scenario, refer to Figure 6 A flow chart is shown. If a passenger finds that the driver has taken a detour after the order is completed and the passenger arrives at the destination, the passenger can file a complaint or feedback through a hotline or online car-hailing client. When the server receives the complaint or feedback, the judgment system, i.e., the trained deep learning framework, is called, and the judgment system performs judgment to obtain the rationality judgment result of the actual trip (it can be understood that the rationality judgment result of the actual trip output by the trained deep learning framework is relatively accurate). Further, based on the rationality judgment result of the actual trip, it is determined whether the driver is responsible. If it is determined that the driver is responsible, a disposal plan is generated, and the corresponding disposal plan is pushed and displayed to the passenger. After the passenger clicks to confirm, a judgment work order is generated to facilitate monitoring and management. If the passenger does not confirm, it means that the passenger does not approve of the disposal plan (for example, the passenger is only compensated 1 yuan, which the passenger thinks is too little and unacceptable). The passenger can continue to complain, and the complaint is transferred to the customer service system at this time, and the manual customer service intervenes to handle it. Or, due to coverage issues, some orders that do not trigger the judgment also directly enter the customer service system and are finally handled by manual customer service. After the processing is completed by manual customer service, the processing results will be recorded to generate new sample data and fed back to the judgment system to continue training the judgment system, so as to continuously improve the performance of the judgment system and make the rationality judgment results of the actual itinerary output by the judgment system more and more accurate.

[0096] The accountability worksheet includes: the rationality judgment result of the actual trip (for example, whether the driver took a detour or not), the disposal plan (for example, how much the passenger should be compensated), the actual departure time, the actual driving route, the actual driving time, etc. Furthermore, the accountability worksheet can be used as new sample data to continue training the deep learning framework, so that the accuracy of the deep learning framework is continuously improved.

[0097] Figure 6The "monitoring and management" in this article refers to returning the data that passengers approve and disapprove of as sample data to the accountability system to continue training the accountability system.

[0098] The method for confirming the rationality of the driving route provided by the embodiment of the present disclosure, by combining the future itinerary prediction technology, estimates the navigation route and travel time from the starting point to the destination at multiple time points before and after the actual departure time, and uses the navigation route and travel time as the input of the accountability system. By using the deep learning framework, the temporal nature of the estimated navigation routes and travel times is effectively utilized to more objectively and effectively mine the information needed for judgment and reference, and further provide the order information and user portrait information to the accountability system, so that the input of the accountability system is more real, more effective and more abundant, thereby obtaining a more accurate judgment result of the rationality of the actual itinerary. In addition, the accountability system immediately conducts objective and timely judgments when passengers give feedback and complaints, which can effectively improve user experience and user churn.

[0099] Figure 7 The schematic diagram of the structure of the device for confirming the rationality of the driving route provided by the embodiment of the present disclosure is as follows: The device for confirming the rationality of the driving route provided by the embodiment of the present disclosure can execute the processing flow provided by the method for confirming the rationality of the driving route.

[0100] like Figure 7 As shown, the device 700 for confirming the rationality of a driving route includes: an acquisition module 710 , an estimation module 720 and a determination module 730 .

[0101] Among them, the acquisition module 710 is used to obtain the actual travel information from the starting point to the destination; the estimation module 720 is used to estimate the navigation route and travel time from the starting point to the destination at a preset time based on the actual travel information; the determination module 730 is used to determine the rationality judgment result of the actual travel by at least using the navigation route and travel time as input features of the deep learning framework when the actual travel information is complained.

[0102] Optionally, the actual itinerary information includes at least one of the following: at least one of an order generation time corresponding to the actual itinerary and an actual departure time corresponding to the actual itinerary.

[0103] Optionally, the estimation module 720 includes: a determination unit, used to determine one or more preset times before the order generation time or the actual departure time; and / or determine one or more preset times after the order generation time or the actual departure time; an estimation unit, used to obtain the navigation route and travel time from the starting point to the destination at the preset time through future itinerary estimation for each of the more than one preset times.

[0104] Optionally, the deep learning framework includes a bidirectional long short-term memory network and a multilayer perceptron; the actual travel information includes the actual driving route; the determination module 730 includes: a first acquisition unit, used to obtain multi-dimensional information based on the actual driving route and the navigation route; a second acquisition unit, used to use the multi-dimensional information and travel time as input features of the bidirectional long short-term memory network to obtain output features of the bidirectional long short-term memory network; a determination unit, used to use at least one of the order information and user portrait information corresponding to the actual travel, and the output features of the bidirectional long short-term memory network as input features of the multilayer perceptron to determine the rationality judgment result of the actual travel.

[0105] Optionally, the multi-dimensional information includes at least one of the following:

[0106] The route length of the navigation route, the length difference between the navigation route and the actual driving route, the number of overlapping road units in the navigation route and the actual driving route, the sum of the lengths of the overlapping road units, the number of non-overlapping road units in the navigation route and the actual driving route, and the sum of the lengths of the non-overlapping road units.

[0107] Optionally, the bidirectional LSTM network includes an attention mechanism layer.

[0108] Optionally, the device 700 for confirming the rationality of the driving route also includes: a first generation module, which is used to generate a treatment plan based on the rationality judgment result of the actual trip after determining the rationality judgment result of the actual trip using at least the navigation route and the travel time as input features of the deep learning framework. A push module, which is used to push the treatment plan to the passenger terminal. A second generation module, which is used to generate sample data when receiving confirmation information on the treatment plan returned by the passenger terminal, and the sample data includes actual trip information. A training module, which is used to train the deep learning framework based on the sample data.

[0109] Figure 7 The device for confirming the rationality of the driving route in the illustrated embodiment can be used to execute the technical solution of the above-mentioned method embodiment. Its implementation principle and technical effect are similar and will not be repeated here.

[0110] The internal functions and structure of the device for confirming the rationality of a driving route are described above. The device can be implemented as an electronic device. Figure 8 This is a schematic diagram of the structure of an electronic device embodiment provided by the present disclosure. Figure 8 As shown, the electronic device includes a memory 151 and a processor 152 .

[0111] The memory 151 is used to store programs. In addition to the above programs, the memory 151 can also be configured to store various other data to support operations on the electronic device. Examples of such data include instructions for any application or method operating on the electronic device, contact data, phone book data, messages, pictures, videos, etc.

[0112] The memory 151 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0113] The processor 152 is coupled to the memory 151 and executes the program stored in the memory 151 to:

[0114] Acquire actual travel information from a starting point to a destination; based on the actual travel information, estimate a navigation route and travel time from the starting point to the destination at a preset time; when a complaint is made about the actual travel information, use at least the navigation route and the travel time as input features of a deep learning framework to determine a rationality judgment result of the actual travel.

[0115] Further, if Figure 8 As shown, the electronic device may also include: a communication component 153, a power component 154, an audio component 155, a display 156 and other components. Figure 8 Only some components are shown schematically, which does not mean that the electronic device only includes Figure 8 Components shown.

[0116] The communication component 153 is configured to facilitate wired or wireless communication between the electronic device and other devices. The electronic device can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 153 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 153 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0117] The power supply component 154 provides power to various components of the electronic device. The power supply component 154 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device.

[0118] The audio component 155 is configured to output and / or input audio signals. For example, the audio component 155 includes a microphone (MIC), and when the electronic device is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode, the microphone is configured to receive an external audio signal. The received audio signal may be further stored in the memory 151 or sent via the communication component 153. In some embodiments, the audio component 155 also includes a speaker for outputting audio signals.

[0119] The display 156 includes a screen, which may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.

[0120] In addition, an embodiment of the present disclosure further provides a computer-readable storage medium on which a computer program is stored, and the computer program is executed by a processor to implement the method for confirming the rationality of the driving route described in the above embodiment.

[0121] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article 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, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0122] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for confirming the rationality of a driving route, wherein: The method comprises: Get the actual travel information from the starting point to the destination; Acquire a preset time according to the actual travel information, and estimate the navigation route and travel time from the starting point to the destination at the preset time; When a complaint is made about the actual trip information, at least the navigation route and the travel time are used as input features of a deep learning framework to determine a rationality judgment result of the actual trip.

2. The method according to claim 1, wherein: The actual itinerary information includes at least one of an order generation time corresponding to the actual itinerary and an actual departure time corresponding to the actual itinerary.

3. The method according to claim 2, wherein: According to the actual travel information, estimating the navigation route and travel time from the starting point to the destination at a preset time, including: Determine one or more preset times before the order generation time or the actual departure time; and / or Determine one or more preset times after the order generation time or the actual departure time; For each of the more than one preset times, a navigation route and travel time from the starting point to the destination at the preset time are obtained through future travel prediction.

4. The method according to claim 1, wherein: The deep learning framework includes a bidirectional long short-term memory network and a multi-layer perceptron; The actual travel information includes the actual driving route; At least the navigation route and the travel time are used as input features of a deep learning framework to determine a judgment result, including: Obtaining multi-dimensional information according to the actual driving route and the navigation route; Using the multi-dimensional information and the travel time as input features of the bidirectional long short-term memory network to obtain output features of the bidirectional long short-term memory network; At least one of the order information and user portrait information corresponding to the actual itinerary and the output features of the bidirectional long short-term memory network are used as input features of the multilayer perceptron to determine the judgment result.

5. The method according to claim 4, characterized in that The multi-dimensional information includes at least one of the following: The route length of the navigation route, the difference in length between the navigation route and the actual driving route, the number of overlapping road units between the navigation route and the actual driving route, the sum of the lengths of the overlapping road units, the number of non-overlapping road units between the navigation route and the actual driving route, and the sum of the lengths of the non-overlapping road units.

6. The method according to claim 4, characterized in that The bidirectional long short-term memory network includes an attention mechanism layer.

7. The method according to claim 1, wherein: At least the navigation route and the travel time are used as input features of the deep learning framework, and after determining the judgment result, the method further includes: Generate a disposal plan based on the judgment result; Pushing the disposal plan to the passenger terminal; When receiving confirmation information of the handling plan returned by the passenger terminal, generating sample data, the sample data including the actual trip information; The deep learning framework is trained according to the sample data.

8. A device for confirming the rationality of a driving route, wherein: include: The acquisition module is used to obtain the actual travel information from the starting point to the destination; An estimation module, used to obtain a preset time according to the actual travel information, and estimate the navigation route and travel time from the starting point to the destination at the preset time; A determination module is used to determine the rationality judgment result of the actual trip by using at least the navigation route and the travel time as input features of the deep learning framework when the actual trip information is complained about.

9. An electronic device, wherein: include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and is configured to be executed by the processor to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images