Driver intent information determination method, client device, server, and system

Abstract

The application provides a driver intention information determination method, a client device, a server and a system, and relates to the technical field of cargo transportation. The method comprises the following steps: when detecting an intention operation behavior of a driver, the client device acquires real-time intention behavior characteristics in a preset time period; whether the intention behavior of the driver changes is determined based on the real-time intention behavior characteristics and historical intention information of the driver; intention recalculation indication information is sent to the server in the case where the intention behavior of the driver changes; and the server re-determines target intention information of the driver based on the intention recalculation indication information. The scheme of the application can timely perceive and respond to real-time intention changes of the driver, and improve the efficiency of cargo source and driver matching.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of goods transportation, and in particular to a driver intention information determination method, a client device, a server and a system. BACKGROUND

[0002] In a logistics transportation service platform, the intention (such as the destination, the departure place, the route, etc.) of a driver is a core basis for matching a cargo source with a driver's capacity, and directly affects the cargo source transaction rate, the driver's experience and the cargo transportation efficiency, etc. With the development of precise matching and other businesses, higher requirements are put forward for the accuracy, real-time performance and calculation efficiency of driver intention recognition.

[0003] In related technologies, driver intention recognition relies on a server-side timed polling method for calculation. Specifically, the server side predicts the intention information such as the destination or route that the driver is likely to transact according to the historical behavior data of the driver every interval of a fixed frequency control time window, and uses the intention information in search recommendation, subscription push and other scenarios. This method cannot timely perceive the intention changes of the driver in a short period of time, and is prone to the situation that the recommended results do not match the real intention of the driver, affecting the efficiency of matching the cargo source with the driver. SUMMARY

[0004] The present application provides a driver intention information determination method, a client device, a server and a system to solve the problem that the intention changes of the driver in a short period of time cannot be timely perceived in related technologies, thereby affecting the efficiency of matching the cargo source with the driver.

[0005] In a first aspect, the present application provides a driver intention information determination method applied to a client device, which comprises:

[0006] In the case that the intention operation behavior of the driver is detected, real-time intention behavior features in a preset time period are acquired;

[0007] Based on the real-time intention behavior features and the historical intention information of the driver, it is determined whether the intention behavior of the driver has changed;

[0008] In the case that the intention behavior of the driver has changed, intention recalculation indication information is sent to a server, so that the server re-determines the target intention information of the driver based on the intention recalculation indication information.

[0009] Optionally, the determination of whether the intention behavior of the driver has changed based on the real-time intention behavior features and the historical intention information of the driver comprises:

[0010] Based on the real-time intention behavior features and the driver portrait features of the driver, initial intention information of the driver is determined;

[0011] determine whether the driver's intended behavior changes based on the initial intention information and the driver's historical intention information.

[0012] Optionally, the determining whether the driver's intended behavior changes based on the initial intention information and the driver's historical intention information comprises:

[0013] determining a correlation index value between the initial intention information and the driver's historical intention information;

[0014] in a case where the correlation index value is less than a preset index threshold, determining that the driver's intended behavior changes;

[0015] in a case where the correlation index value is greater than or equal to the preset index threshold, determining that the driver's intended behavior does not change.

[0016] Optionally, the determining the initial intention information of the driver based on the real-time intention behavior feature and the driver portrait feature comprises:

[0017] obtaining the driver portrait feature of the driver;

[0018] inputting the real-time intention behavior feature and the driver portrait feature into a first intention recognition model to obtain the initial intention information of the driver output by the first intention recognition model;

[0019] wherein the first intention recognition model is used for identifying the intention of the driver based on the real-time intention behavior feature and the driver portrait feature; and the first intention recognition model is obtained by training a first initial intention recognition model based on first sample real-time intention behavior features, first sample driver portrait features, and corresponding first label data.

[0020] In a second aspect, the present application provides a driver intention information determination method applied to a server, comprising:

[0021] in a case where the intention recalculation indication information sent by the client device is received, determining the target intention information of the driver again based on the intention recalculation indication information;

[0022] wherein the intention recalculation indication information is sent by the client device in a case where the driver's intention operation behavior is detected, and the driver's intended behavior is determined to change based on real-time intention behavior features in a preset time period and historical intention information of the driver.

[0023] Optionally, the real-time intention behavior feature is included in the intention recalculation indication information; and the target intention information of the driver is determined based on the intention recalculation indication information, comprising:

[0024] The target attribute feature of the driver is acquired, and the target attribute feature at least includes the driver portrait feature and the driver historical behavior feature of the driver;

[0025] The intention of the driver is determined based on the real-time intention behavior feature and the target attribute feature, and the target intention information is obtained.

[0026] Optionally, the intention of the driver is determined based on the real-time intention behavior feature and the target attribute feature, and the target intention information is obtained, comprising:

[0027] The real-time intention behavior feature and the target attribute feature are input into a second intention recognition model, and the target intention information of the driver output by the second intention recognition model is obtained;

[0028] The second intention recognition model is used for recognizing the intention of the driver based on the real-time intention behavior feature and the target attribute feature; the second intention recognition model is obtained by training a second initial intention recognition model based on a second sample real-time intention behavior feature, a sample target attribute feature and a corresponding second label data; and the sample target attribute feature at least includes a second sample driver portrait feature and a sample driver historical behavior feature.

[0029] In a third aspect, a client device is provided, which comprises a first memory and a first processor. The first memory stores a computer program capable of running on the first processor. When the first processor executes the computer program, the steps of the driver intention information determination method according to any one of the first aspect are implemented.

[0030] In a fourth aspect, a server is provided, which comprises a second memory and a second processor. The second memory stores a computer program capable of running on the second processor. When the second processor executes the computer program, the steps of the driver intention information determination method according to any one of the second aspect are implemented.

[0031] In a fifth aspect, a driver intention information determination system is provided, which comprises the client device according to the third aspect and the server according to the fourth aspect. The client device and the server are in communication connection.

[0032] The driver intention information determination method, the client device, the server and the system provided by the application, the client device detects the intention operation behavior of the driver, obtains the real-time intention behavior characteristics in the preset time period, determines whether the intention behavior of the driver changes based on the real-time intention behavior characteristics and the historical intention information of the driver, and sends the intention recalculation indication information to the server in the case that the intention behavior of the driver changes. The server re-determines the target intention information of the driver based on the intention recalculation indication information after receiving the intention recalculation indication information. In this way, the real-time detection of the intention behavior of the driver by the client device and the re-determination of the intention information of the driver by the server when the intention behavior changes can timely perceive and respond to the real-time intention change of the driver, improve the real-time performance and accuracy of the driver intention recognition, make the recommended result determined based on the driver intention information more consistent with the real intention of the driver, and further improve the efficiency of the matching of the cargo source and the driver. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 One of the flowcharts of the driver intention information determination method provided by the embodiments of the application;

[0034] Figure 2 The second flowchart of the driver intention information determination method provided by the embodiments of the application;

[0035] Figure 3 The third flowchart of the driver intention information determination method provided by the embodiments of the application;

[0036] Figure 4 The structure diagram of the driver intention information determination system provided by the embodiments of the application. DETAILED DESCRIPTION

[0037] In the application, "at least one" means one or more, and "multiple" means two or more. The "and / or" describes the association relationship of the associated objects, which means that there can be three kinds of relationships, for example, A and / or B, which can represent the following cases: A exists alone, A and B exist together, and B exists alone. Wherein A and B can be singular or plural. The character " / " generally represents an "or" relationship between the front and rear associated objects. "At least one of the following" or similar expressions means any combination of these items, including any combination of single item or multiple items. For example, at least one of a, b or c alone, which can represent: a alone, b alone, c alone, combination of a and b, combination of a and c, combination of b and c, or combination of a, b and c. Wherein a, b and c can be single or multiple. In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance.

[0038] The terms "connected" and "connection" should be interpreted broadly, for example, the "connection" of a circuit structure can be not only a physical connection, but also an electrical connection or a signal connection. For example, it can be a direct connection, that is, a physical connection, or an indirect connection through at least one element, as long as the circuit is connected. It can also be an internal connection of two elements. In addition to electrical connection, signal connection can also be achieved through media such as radio waves. For those skilled in the art, the specific meaning of the above terms in this application can be understood according to the specific circumstances.

[0039] In a logistics transportation service platform, the intention of a driver (such as a destination, a departure place, a route, etc.) is the core basis for matching a cargo source with a driver's capacity, and directly affects the cargo source transaction rate, the driver's experience, and the cargo transportation efficiency, etc. With the development of precise matching and other businesses, higher requirements are put forward for the accuracy, real-time performance, and computing efficiency of driver intention recognition.

[0040] In related technologies, the recognition of driver intention depends on the server-side timed polling method for calculation. That is, the server side determines the intention information of the driver according to a fixed frequency control time window, and at intervals, such as every 5 minutes, predicts the possible transaction destination or route of the driver by analyzing the historical behavior data of the driver, such as historical transactions, clicks, and exposures, and uses the intention information for search recommendation, subscription push, and other scenarios. Since this method determines the intention information according to a fixed frequency control time window, it cannot timely perceive the changes in the driver's intention within a short period of time, and is prone to mismatches between the recommended results and the driver's real intention, thereby affecting the efficiency of matching the cargo source with the driver. Moreover, according to this method, whether the driver's intention changes or not, the intention information will be determined at intervals. For the case where the driver's intention does not change in multiple consecutive requests, repeated calculations will be performed, resulting in waste of computing resources.

[0041] Therefore, the embodiments of the present application provide a driver intention information determination method, which can detect whether the driver's intention behavior has changed through a client device, and only trigger the server to determine the driver's intention information again when the driver's intention behavior has changed.

[0042] The embodiments of the present application will be described in detail below. Figures 1-3 The driver intention information determination method provided by the embodiments of the present application will be described in detail.

[0043] Figure 1Fig. 1 shows a flowchart of a driver intention information determination method provided by an embodiment of the present application. The driver intention information determination method can be applied to a client device or a first driver intention information determination apparatus provided in the client device. The first driver intention information determination apparatus can be implemented by software, hardware or a combination of both. Figure 1 As shown in Fig. 1, the driver intention information determination method can include steps 110-130.

[0044] Step 110: In a case where a driver's intention operation behavior is detected, real-time intention behavior features in a preset time period are acquired.

[0045] The driver can perform an intention operation on a logistics transportation service platform through a client device. The intention operation can include clicking a vote source, searching a route or refreshing a page, but is not limited thereto. The client device can detect the driver's intention operation behavior, and in a case where the intention operation behavior is detected, real-time intention behavior features in a preset time period starting from a current time when the intention operation behavior is detected can be acquired. For example, route click times and route exposure click rates in N (N is a positive integer) minutes starting from the current time can be acquired as real-time intention behavior features.

[0046] Step 120: Based on the real-time intention behavior features and historical intention information of the driver, it is determined whether the driver's intention behavior has changed.

[0047] After the client device acquires the real-time intention behavior features, historical intention information of the driver can be acquired, and then based on the real-time intention behavior features of the driver and the historical intention information of the driver, it is determined whether the driver's intention behavior has changed.

[0048] The historical intention information of the driver can include at least one of historical destinations, historical departure locations and historical routes, but is not limited thereto.

[0049] For example, the historical intention information of the driver can be intention information of the driver determined by the client device when the driver performed an intention operation behavior last time. The historical intention information can be cached in a cache device of the client device, and the client device can directly read the historical intention information of the driver from the cache device.

[0050] For example, after the client device acquires the real-time intention behavior features, the client device can determine initial intention information of the driver based on the real-time intention behavior features, and compare the initial intention information with the historical intention information of the driver, and determine whether the driver's intention behavior has changed according to a comparison result.

[0051] Specifically, step 120 determines whether the driver's intention behavior changes based on the real-time intention behavior feature and the historical intention information of the driver, which can be implemented through steps 121-122 as follows.

[0052] Step 121: determining the initial intention information of the driver based on the real-time intention behavior feature and the driver portrait feature of the driver.

[0053] The driver portrait feature of the driver can include at least one of historical frequently traveled routes, historical frequently traveled cities, and historical frequently carried cargo source weights, but is not limited thereto. The driver portrait feature can depict the characteristics of the freight transportation portrait of the driver.

[0054] The initial intention information can include at least one dimension of intention information such as initial departure location, initial destination, and initial route. For each dimension of intention, the intentions can be sorted in descending order of intention score, and the top pre-set number of intentions in the sorting can be taken as the intention information of the dimension of intention. The pre-set number can be designed according to actual needs, such as 5, 8, or 10, etc.

[0055] For example, after the client device obtains the real-time intention behavior feature of the driver, the driver portrait feature of the driver can be obtained, and then the intention preference feature of the driver can be determined based on the driver portrait feature of the driver. The real-time intention behavior feature of the driver and the intention preference feature are compared and fused to obtain the initial intention information of the driver.

[0056] Alternatively, the client device can determine the initial intention information of the driver based on the real-time intention behavior feature of the driver and the driver portrait feature by using the trained first intention recognition model.

[0057] Specifically, step 121 determines the initial intention information of the driver based on the real-time intention behavior feature and the driver portrait feature of the driver, which can include:

[0058] Obtaining the driver portrait feature of the driver; inputting the real-time intention behavior feature and the driver portrait feature into the first intention recognition model to obtain the initial intention information of the driver output by the first intention recognition model; wherein the first intention recognition model is used to identify the intention of the driver based on the real-time intention behavior feature and the driver portrait feature; the first intention recognition model is obtained by training the first initial intention recognition model based on the first sample real-time intention behavior feature, the first sample driver portrait feature, and the corresponding first label data.

[0059] The driver portrait feature of the driver can be transparently transmitted to the client device by the server. The server can determine the driver portrait feature of the driver according to historical orders of the driver. When the driver logs in the logistics transportation service platform through the client device, the server can transparently transmit the driver portrait feature of the driver to the client device. The client device can cache the driver portrait feature of the driver.

[0060] After the client device obtains the real-time intention behavior feature of the driver, the client device can obtain the driver portrait feature of the driver from the cache device, and then input the real-time intention behavior feature and the driver portrait feature into the first intention recognition model to perform intention recognition on the driver by using the first intention recognition model, to obtain initial intention information of the driver.

[0061] For example, the first initial intention recognition model can be trained by using the client device or other model training devices. If the other model training devices are used for training, the trained first intention recognition model can be deployed on the client device after the training is completed. The other model training devices can include electronic devices such as computers, notebook computers, or tablet computers, but are not limited thereto.

[0062] The client device or the other model training devices can collect a large amount of first sample driver historical data from the logistics transportation service platform, obtain first sample real-time intention behavior features and first sample driver portrait features from the first sample driver historical data, and label whether a one-vote order is transacted for the first sample driver historical data to obtain first label data. Then, the first sample real-time intention behavior features and the first sample driver portrait features are input into the first initial intention recognition model to obtain a first sample intention ranking result output by the first initial intention recognition model. Then, the model parameters of the first initial intention recognition model are adjusted based on the first sample intention ranking result and the first label data, until the first initial intention recognition model converges, to obtain the trained first intention recognition model.

[0063] The first initial intention recognition model and the first intention recognition model can be a Mixture of Experts (MoE) model or a Multi-gate Mixture-of-Experts (MMOE) model. These models can be lightweight neural network models.

[0064] The trained first intention recognition model can be deployed on the client device, which can support low-latency response.

[0065] Step 122: Determine whether the intention behavior of the driver changes based on the initial intention information and historical intention information of the driver.

[0066] After obtaining the initial intention information of the driver, the client device can compare the initial intention information of the driver with historical intention information of the driver, and determine whether the intention behavior of the driver has changed according to a comparison result.

[0067] For example, the step 122 of determining whether the intention behavior of the driver has changed based on the initial intention information and the historical intention information of the driver can include:

[0068] determining a correlation index value between the initial intention information and the historical intention information of the driver; determining that the intention behavior of the driver has changed in a case where the correlation index value is less than a preset index threshold; and determining that the intention behavior of the driver has not changed in a case where the correlation index value is greater than or equal to the preset index threshold.

[0069] The correlation index value is used to represent the correlation between the initial intention information and the historical intention information, and the greater the correlation index value, the more relevant the initial intention information and the historical intention information. The preset index threshold can be set according to experience or actual needs, for example, set to 0.6, etc.

[0070] In a case where the correlation index value is less than the preset index threshold, it indicates that the correlation between the initial intention information and the historical intention information is low, and it can be considered that they are not relevant, and then it can be determined that the intention behavior of the driver has changed. In a case where the correlation index value is greater than or equal to the preset index threshold, it indicates that the correlation between the initial intention information and the historical intention information is high, and it can be considered that the initial intention information and the historical intention information are almost the same, and then it can be determined that the intention behavior of the driver has not changed.

[0071] For example, the correlation index value can include a Kendall tau coefficient or a Spearman rank correlation coefficient, etc.

[0072] For example, the historical intention information of the driver can be the intention information of the driver determined by the client device when the driver performed an intention operation behavior last time.

[0073] For example, the driver clicks a vote A place to B place and then to C place at the first time T1. After the client device detects the click operation behavior, the real-time intention behavior features in the preset time period can be obtained based on the click operation behavior, and then the initial intention information of the driver is determined according to step 121. The initial intention information includes the destination intention ranking score of the driver, such as A place 0.9, C place 0.6, and B place 0.4. The initial intention information can be cached in the client device as the historical intention information of the driver for the next intention operation. For example, the driver clicks a vote from A place to C place and then to B place at the second time T2 after the first time T1. After the client device detects the click operation behavior, the initial intention information of the driver can be determined by the same method. The initial intention information includes the destination intention ranking score of the driver, such as B place 0.7, A place 0.6, and C place 0.56. At this time, the client device can calculate the relevance of the initial intention information at the second time T2 and the initial intention information at the first time T1 as historical intention information to determine whether the intention operation behavior of the driver at the second time T2 changes.

[0074] In this way, the relevance index value can quantify the ranking difference between the current intention and the historical intention on the client side, accurately identify the real intention change of the driver, and effectively alleviate the missed trigger or false trigger of intention calculation.

[0075] Step 130: In the case where the intention behavior of the driver changes, intention recalculation indication information is sent to the server to make the server determine the target intention information of the driver based on the intention recalculation indication information.

[0076] The intention recalculation indication information can be used to instruct the server to determine the intention information of the driver again to obtain the target intention information.

[0077] For example, the intention recalculation indication information can include the real-time intention behavior features of the driver, such as a driver identifier and corresponding real-time intention behavior features. After the server receives the intention recalculation indication information, the intention information of the driver can be determined again based on the real-time intention behavior features of the driver.

[0078] In this way, the server can only start the program of determining the intention information of the driver when the client device detects that the intention behavior of the driver changes. Not only can the real-time intention change of the driver be responded in time, but also compared with the intention information updating method according to the fixed frequency control time window in the prior art, the problem of wasting computing resources caused by frequent triggering of redundant calculation when the intention behavior of the driver does not change is avoided, and the computing resources are saved.

[0079] The driver intent information determination method provided in this application involves a client device detecting a driver's intent operation behavior, acquiring real-time intent behavior characteristics within a preset time period, and determining whether the driver's intent behavior has changed based on the real-time intent behavior characteristics and the driver's historical intent information. If the driver's intent behavior has changed, the client device sends intent recalculation instruction information to the server, instructing the server to re-determine the driver's target intent information based on the intent recalculation instruction information. In this way, by having the client device detect changes in the driver's intent behavior in real time and trigger the server to re-determine the driver's intent information when changes occur, the method can promptly perceive and respond to real-time changes in the driver's intent, improving the real-time performance and accuracy of driver intent recognition. This makes the recommendation results determined by the server based on the driver's intent information more consistent with the driver's true intent, thereby improving the efficiency of matching cargo with drivers.

[0080] based on Figure 1 The driver intent information determination method in the corresponding embodiment, Figure 2 This illustration shows a second flowchart of a driver intent information determination method provided in an embodiment of this application. This method can be applied to a server, which may include a standalone server, a virtual server, or a cluster server. The method can also be applied to a second driver intent information determination device installed on the server. This second driver intent information determination device can be implemented through software, hardware, or a combination of both. (Refer to...) Figure 2 As shown, the method for determining driver intent information may include the following step 210.

[0081] Step 210: Upon receiving the intent recalculation instruction information sent by the client device, re-determine the driver's target intent information based on the intent recalculation instruction information.

[0082] Among them, the intent recalculation indication information is sent by the client device when it detects the driver's intent operation behavior and determines that the driver's intent behavior has changed based on the real-time intent behavior characteristics within a preset time period and the driver's historical intent information.

[0083] For example, the intent recalculation instruction information may include the driver's real-time intent behavior characteristics, such as a driver identifier and corresponding real-time intent behavior characteristics. After receiving the intent recalculation instruction information, the server can re-determine the driver's intent information based on the driver's real-time intent behavior characteristics.

[0084] Specifically, step 210, which re-determines the driver's target intent information based on the intent recalculation instruction information, can be achieved through the following steps 211 to 212.

[0085] Step 211: Obtain target attribute features of the driver.

[0086] The target attribute features at least include driver portrait features and historical behavior features of the driver. The driver portrait features of the driver can include at least one of historical frequently traveled routes, historical frequently traveled cities, and historical frequently carried cargo weights, but are not limited thereto. The driver portrait features can depict the freight transportation portrait characteristics of the driver. The historical behavior features of the driver are used to represent the historical behavior of the driver, which can include at least one of historical freight source click operations, historical route search operations, and historical route subscription operations, but are not limited thereto.

[0087] For example, the intent recalculation indication information can include a driver identifier of the driver, and the server can obtain the target attribute features corresponding to the driver identifier.

[0088] For example, the server can determine the driver portrait features and the historical behavior features of the driver according to historical orders of the driver. For example, the server can extract the driver portrait features and the historical behavior features of the driver from historical orders of the driver within a preset historical time period. After determining the driver portrait features of the driver, the server can cache the driver portrait features, for example, as a corresponding relationship between the driver identifier and the driver portrait features.

[0089] For example, when the driver logs in to the logistics transportation service platform through the client device, the server can also transmit the driver portrait features of the driver to the client device, and the client device can cache the driver portrait features of the driver.

[0090] Step 212: Redetermine the intent of the driver based on the real-time intent behavior features and the target attribute features, and obtain target intent information.

[0091] After the server receives the intent recalculation indication information sent by the client device, the server can obtain the real-time intent behavior features of the driver from the intent recalculation indication information. After the server obtains the real-time intent behavior features and the target attribute features of the driver, the server can redetermine the intent of the driver based on the real-time intent behavior features and the target attribute features, and obtain target intent information.

[0092] The target intent information can include intent information of at least one dimension of the departure location, the destination, and the route. For each dimension of the intent information, the intent information can be sorted according to the scoring value of the intent. For example, the intents of each dimension can be sorted in descending order of the scoring value, and the first preset number of intents are selected as the intent information of the dimension.

[0093] After the server obtains the target intention information, the server can apply the target intention information to a search recommendation, a subscription push, and the like, determine a recommendation list corresponding to the driver based on the target intention information, such as a cargo source recommendation list or a subscription information list, and then send the recommendation list to the client device.

[0094] In an embodiment, the server can determine the intention of the driver again by using the trained second intention recognition model based on the real-time intention behavior feature and the target attribute feature. Specifically, step 212 of determining the intention of the driver again based on the real-time intention behavior feature and the target attribute feature to obtain the target intention information can include: inputting the real-time intention behavior feature and the target attribute feature into the second intention recognition model to obtain the target intention information of the driver output by the second intention recognition model.

[0095] The second intention recognition model is used to identify the intention of the driver based on the real-time intention behavior feature and the target attribute feature; the second intention recognition model is obtained by training the second initial intention recognition model based on the second sample real-time intention behavior feature, the sample target attribute feature, and the corresponding second label data; and the sample target attribute feature at least includes the second sample driver portrait feature and the sample driver historical behavior feature.

[0096] For example, the second initial intention recognition model and the second intention recognition model can be a deep neural network model such as multi-task learning (MTL), or can be a large language model (LLM).

[0097] For example, the second intention recognition model can be a neural network model with higher intention recognition accuracy than the first intention recognition model, and can output an intention ranking result with higher accuracy than the first intention recognition model.

[0098] For example, the second initial intention recognition model can be trained by using a server or other model training device. If other model training devices are used for training, the trained second intention recognition model can be deployed on the server after the training is completed. The other model training devices can include electronic devices such as computers, notebook computers, or tablet computers, but are not limited thereto.

[0099] The server or other model training device can collect a large amount of second sample driver historical data from the logistics transportation service platform, obtain second sample real-time intention behavior features and sample target attribute features from the second sample driver historical data, and label whether a one-vote cargo source is transacted for the second sample driver historical data to obtain second label data, then input the second sample real-time intention behavior features and the sample target attribute features into the second initial intention recognition model to obtain a second sample intention ranking result output by the second initial intention recognition model, and then adjust model parameters of the second initial intention recognition model based on the second sample intention ranking result and the second label data until the second initial intention recognition model converges to obtain a trained second intention recognition model.

[0100] The sample target attribute features at least include sample driver portrait features and sample driver historical behavior features of the sample driver.

[0101] The driver intention information determination method provided in the embodiments of the present application, the server re-determines the target intention information of the driver based on the intention recalculation indication information after receiving the intention recalculation indication information sent by the client device when detecting the intention operation behavior of the driver based on the real-time intention behavior features within the preset time period and the historical intention information of the driver in the case of determining that the intention behavior of the driver changes. In this way, by detecting whether the intention behavior of the driver changes in real time through the client device, and triggering the server to re-determine the intention information of the driver when the change occurs, the real-time intention change of the driver can be perceived and responded in time, and the real-time and accuracy of driver intention recognition are improved, so that the recommended result determined by the server based on the driver intention information is more consistent with the real intention of the driver, and the efficiency of matching of the cargo source and the driver is improved.

[0102] Based on Figure 1 and Figure 2 the driver intention information determination method of the corresponding embodiments, Figure 3 Fig. 3 shows a flowchart of a driver intention information determination method provided in the embodiments of the present application. The driver intention information determination method can be applied to a driver intention information determination system provided in the embodiments of the present application. The driver intention information determination system can include a client device and a server. As shown in Fig. 1, the driver intention information determination method can include the following steps 101-107. Figure 3

[0103] Step 301: The client device obtains real-time intention behavior features within a preset time period in the case of detecting an intention operation behavior of a driver.

[0104] Specifically, step 301 corresponds to step 110 described above, and the specific implementation process can refer to the description of step 110 described above, which will not be described here again. ​

[0105] Step 302: The client device determines initial intention information of the driver based on the real-time intention behavior feature and the driver portrait feature of the driver.

[0106] Specifically, step 302 corresponds to step 121 described above, and the specific implementation process can refer to the description of step 121 described above, which will not be described here.

[0107] Step 303: The client device determines a correlation index value between the initial intention information and the historical intention information of the driver.

[0108] The correlation index value is used to represent the correlation between the initial intention information and the historical intention information, and the greater the correlation index value, the more relevant the initial intention information and the historical intention information.

[0109] For example, the correlation index value can include Kendall tau coefficient or Spearman rank correlation coefficient, etc.

[0110] For example, the historical intention information of the driver can be the intention information of the driver determined by the client device when the driver performed the intention operation behavior last time.

[0111] Step 304: The client device determines whether the correlation index value is less than a preset index threshold.

[0112] The preset index threshold can be set according to experience or actual needs, such as 0.6, etc.

[0113] In the case where it is determined that the correlation index value is less than the preset index threshold, it indicates that the correlation between the initial intention information and the historical intention information is low, and it can be considered that they are not relevant, then the client device determines that the intention behavior of the driver has changed, at this time, step 305 can be executed; otherwise, it indicates that the correlation between the initial intention information and the historical intention information is high, and it can be considered that the initial intention information and the historical intention information are almost the same, then the client device determines that the intention behavior of the driver has not changed, and step 307 is executed.

[0114] Step 305: The client device sends intention recalculation indication information to the server.

[0115] The intention recalculation indication information can be used to instruct the server to determine the intention information of the driver again to obtain target intention information.

[0116] For example, the intention recalculation indication information can include the real-time intention behavior feature of the driver, such as the driver identifier and the corresponding real-time intention behavior feature.

[0117] Step 306: The server determines the target intention information of the driver based on the intention recalculation indication information.

[0118] After the server receives the intention recalculation indication information, the target intention information of the driver can be determined again based on the intention recalculation indication information.

[0119] Specifically, step 306 can correspond to step 210 described above, and the specific implementation process can refer to the description of step 210 above, which will not be described here.

[0120] Step 307: The server keeps the target intention information of the driver unchanged.

[0121] If the client device determines that the intention behavior of the driver has not changed, the client device will not send the intention recalculation indication information to the server, and at this time the server will not recalculate the intention information of the driver, that is, the server keeps the current target intention information of the driver unchanged. In this way, the server can only recalculate the intention information when the intention behavior of the driver changes.

[0122] The driver intention information determination method provided by the embodiment of the application, in the case that the client device detects the intention operation behavior of the driver, the real-time intention behavior characteristics of the driver in a preset time period are obtained through the client device, whether the intention behavior of the driver changes is determined based on the real-time intention behavior characteristics and the historical intention information of the driver, in the case that the intention behavior of the driver changes, the intention recalculation indication information is sent to the server, and after the server receives the intention recalculation indication information, the target intention information of the driver is determined again based on the intention recalculation indication information. In this way, through the client device, whether the intention behavior of the driver changes is detected in real time, and the server determines the intention information of the driver again when the intention behavior changes, the collaborative mechanism of "client-side triggering + server-side recalculation" is used, which breaks the pure server-end dominant mode in the related art, and realizes the upgrade of the driver intention recognition from "passive polling" to "active perception" and the mode upgrade from "timed full calculation" to "change-driven and on-demand triggering". On the one hand, the real-time intention change of the driver can be perceived and responded in time, the real-time and accuracy of the driver intention recognition are improved, the recommendation result determined by the server based on the driver intention information is more consistent with the real intention of the driver, and thus the efficiency of the matching of the cargo source and the driver is improved. On the other hand, since the server determines the intention information of the driver again only in the case that the intention behavior of the driver changes, compared with the polling mode in the related art, redundant invalid triggering is avoided, and the computing resources are effectively saved.

[0123] The embodiment of the application also provides a client device, which can include a first memory and a first processor, the first memory stores a computer program capable of running on the first processor, and the first processor can implement the method as described above when executing the computer program. Figure 1Corresponding to the steps of the driver intention information determination method of the embodiment, which will not be repeated here.

[0124] The embodiment of the present application further provides a server, comprising a second memory and a second processor, the second memory stores a computer program capable of running on the second processor, and the second processor can realize the method as described above when executing the computer program. Figure 2 Corresponding to the steps of the driver intention information determination method of the embodiment, which will not be repeated here.

[0125] The embodiment of the present application further provides a driver intention information determination system, Figure 3 The structure schematic diagram of the driver intention information determination system provided by the embodiment of the present application is shown, referring to Figure 4 Figure 4 It is shown that the driver intention information determination system can comprise a client device 410 and a server 420, and the client device 410 and the server 420 can be communicatively connected, such as being communicatively connected through the Internet or a mobile communication network, and can communicate data with each other.

[0126] The client device 410 can be the client device described in the above embodiment, and the server 420 can be the server described in the above embodiment.

[0127] Based on the driver intention information determination method described in any of the above embodiments, the embodiment of the present application further provides a computer readable storage medium, for example, a non-transitory computer readable storage medium can be a read-only memory (ReadOnly Memory, ROM), a random access memory (Random Access Memory, RAM), a CD-ROM, a magnetic tape, a floppy disk and an optical data storage device, etc. The storage medium stores computer instructions for executing the driver intention information determination method described in any of the above embodiments, which will not be repeated here.

[0128] Those skilled in the art can understand that all or part of the steps of the above embodiments can be completed by hardware, or by program to instruct related hardware, and the program can be stored in a computer readable storage medium, and the above mentioned storage medium can be a read-only memory, a magnetic disk or an optical disk, etc.

[0129] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. The specification and examples given are exemplary only and the true scope and spirit of the application is indicated by the claims. The specification and examples are to be regarded as illustrative only, the true scope and spirit of the application being indicated by the claims.

Claims

1. A driver intention information determination method characterized by comprising: The driver intention information determination method applied to a client device comprises: In the case of detecting the intention operation behavior of the driver, real-time intention behavior characteristics in a preset time period are acquired; Based on the real-time intention behavior characteristics and the historical intention information of the driver, it is determined whether the intention behavior of the driver has changed; In the case of the intention behavior of the driver changing, intention recalculation indication information is sent to a server, so that the server re-determines the target intention information of the driver based on the intention recalculation indication information; The method comprises: Based on the real-time intention behavior characteristics and the driver portrait characteristics of the driver, initial intention information of the driver is determined; Based on the initial intention information and the historical intention information of the driver, it is determined whether the intention behavior of the driver has changed; The method comprises: The correlation index value between the initial intention information and the historical intention information of the driver is determined; In the case that the correlation index value is less than a preset index threshold, it is determined that the intention behavior of the driver has changed; In the case that the correlation index value is greater than or equal to the preset index threshold, it is determined that the intention behavior of the driver has not changed.

2. The driver intention information determination method according to claim 1, characterized by, The method comprises: The driver portrait characteristics of the driver are acquired; The real-time intention behavior characteristics and the driver portrait characteristics are input into a first intention recognition model to obtain the initial intention information of the driver output by the first intention recognition model; The first intention recognition model is used for recognizing the intention of the driver based on the real-time intention behavior characteristics and the driver portrait characteristics; and the first intention recognition model is obtained by training a first initial intention recognition model based on first sample real-time intention behavior characteristics, first sample driver portrait characteristics and corresponding first label data.

3. A driver intention information determination method characterized by comprising: The driver intention information determination method applied to a server comprises: In the case of receiving intention recalculation indication information sent by a client device, the target intention information of a driver is re-determined based on the intention recalculation indication information; The intention recalculation indication information is sent by the client device in a case where the driver's intention behavior is detected to change based on real-time intention behavior features in a preset time period and historical intention information of the driver.

4. The driver's intention information determination method according to claim 3, characterized by, The intention recalculation indication information includes the real-time intention behavior features. The target intention information of the driver is re-determined based on the intention recalculation indication information, including: Obtaining target attribute features of the driver, the target attribute features at least including driver portrait features and driver historical behavior features of the driver; Re-determining the intention of the driver based on the real-time intention behavior features and the target attribute features to obtain target intention information.

5. The driver intention information determination method according to claim 4, characterized by, The target intention information of the driver is re-determined based on the real-time intention behavior features and the target attribute features, including: Inputting the real-time intention behavior features and the target attribute features into a second intention recognition model to obtain target intention information of the driver output by the second intention recognition model; The second intention recognition model is used to recognize the intention of the driver based on the real-time intention behavior features and the target attribute features; the second intention recognition model is obtained by training a second initial intention recognition model based on second sample real-time intention behavior features, sample target attribute features and corresponding second label data; the sample target attribute features at least include second sample driver portrait features and sample driver historical behavior features.

6. A client device comprising a first memory and a first processor, the first memory storing a computer program executable on the first processor, characterized in that, The first processor executes the computer program to implement the steps of the driver intention information determination method of claim 1 or 2.

7. A server comprising a second memory and a second processor, said second memory storing a computer program executable on said second processor, characterized in that, The second processor executes the computer program to implement the steps of the driver intention information determination method of any one of claims 3 to 5.

8. A driver intention information determination system characterized by comprising: The client device of claim 6 and the server of claim 7 are communicatively connected.

Images