Recommended method, device, electronic device and storage medium for on-board solution

Abstract

The present application provides a method, device, electronic device and storage medium for recommending a boarding solution. It includes: after a passenger initiates an online car-hailing order on an online car-hailing platform, obtaining the voice call information between the driver and the passenger, as well as the order information of the online car-hailing order; calling a voice recognition model to process the voice call information to obtain the current location description information of the driver and the passenger; calling a text sorting model to process the location positioning result and order information of the current location description information to obtain text information in a specified text format; calling a solution output model to process the text information in a specified text format to obtain a boarding solution; according to the pick-up location in the boarding solution and the current location of the driver and the passenger, determine the target pick-up location with the lowest pick-up cost, and recommend the boarding solution corresponding to the target pick-up location to the driver and the passenger. This application can improve the driver and passenger experience and reduce order loss.

Description

Technical Field

[0001] The present application relates to the field of online car-hailing technology, and in particular to a method, device, electronic device and storage medium for recommending a boarding solution. Background Art

[0002] The existing application of large language models in the online car-hailing and travel industries is mainly concentrated in the aspects of intelligent customer service and risk warning, both during the order process and after the order is completed. In these two links, the connection between the driver and the passenger has been formed, the uncertainty of the order is low, and most of the information is already determined, which is a situation where it is easier for the large model to intervene. For example, the large model determines whether the driver has risky driving during the driving process, focusing on preventing accidents in the process of determining the itinerary. In the initial stage of forming an online car-hailing order, especially the driver-passenger meeting stage that is prone to disputes and order loss, it is often highly dependent on the personal efforts and communication of both the driver and the passenger, and lacks appropriate means of intervention.

[0003] In the early stages of forming an online car-hailing order, especially when the driver picks up the passenger, it is highly dependent on the personal efforts and communication of both the driver and the passenger to complete the driver-passenger meeting. However, the information in the driver-passenger conversation often cannot be understood or interfered with by a third party. Once the driver and the passenger do not understand or cannot find each other, it is very easy to cause disputes between the driver and the passenger and the loss of orders. Existing technical solutions are rarely used in this link, and there is a lack of means to intervene in the starting point in real time. They can often only intervene in the post-judgment part, which leads to the loss of orders and reduces the experience of the driver and passengers. Summary of the invention

[0004] The technical problem to be solved by the embodiments of the present application is to provide a method, device, electronic device and storage medium for recommending a boarding solution, so as to understand the semantics and location descriptions of both the driver and the passenger through a deep learning model, generate a new starting point and solution to guide the driver and passenger to meet, improve the driver and passenger experience, and reduce order loss.

[0005] In a first aspect, an embodiment of the present application provides a method for recommending a boarding solution, the method comprising:

[0006] After the passenger places an online car-hailing order on the online car-hailing platform, obtain the voice call information between the driver and the passenger during the telephone call, as well as the order information of the online car-hailing order;

[0007] Invoking a speech recognition model to process the voice call information to obtain current location description information of the driver and the passenger;

[0008] Calling a text arrangement model to process the location positioning result corresponding to the current location description information and the order information to obtain text information in a specified text format;

[0009] Calling the solution output model to process the text information in the specified text format to obtain a boarding solution corresponding to the online car-hailing order;

[0010] According to the pick-up location in the boarding solution and the current locations of the driver and the passenger, a target pick-up location with the minimum pick-up cost among the pick-up locations is determined, and a boarding solution corresponding to the target pick-up location is recommended to the driver and the passenger.

[0011] In a second aspect, an embodiment of the present application provides a device for recommending a boarding solution, the device comprising:

[0012] A voice call information acquisition module is used to obtain the voice call information of the driver and the passenger during the telephone call after the passenger initiates an online car-hailing order on the online car-hailing platform, as well as the order information of the online car-hailing order;

[0013] A location description information acquisition module, used to call a speech recognition model to process the voice call information and obtain current location description information of the driver and the passenger;

[0014] A text information acquisition module, used for invoking a text arrangement model to process the position positioning result corresponding to the current position description information and the order information to obtain text information in a specified text format;

[0015] A boarding solution acquisition module, used to call the solution output model to process the text information in the specified text format to obtain the boarding solution corresponding to the online car-hailing order;

[0016] The boarding solution recommendation module is used to determine the target pick-up location with the lowest pick-up cost among the pick-up locations based on the pick-up location in the boarding solution and the current locations of the driver and the passenger, and recommend the boarding solution corresponding to the target pick-up location to the driver and the passenger.

[0017] In a third aspect, an embodiment of the present application provides an electronic device, including:

[0018] A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for recommending a boarding solution described in any one of the above items is implemented.

[0019] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium. When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute any of the above-mentioned methods for recommending on-board solutions.

[0020] Compared with the prior art, the embodiments of the present application have the following advantages:

[0021] In an embodiment of the present application, after the passenger initiates an online car-hailing order on the online car-hailing platform, the voice call information of the driver and the passenger during the telephone call and the order information of the online car-hailing order are obtained. The voice recognition model is called to process the voice call information to obtain the current location description information of the driver and the passenger. The text sorting model is called to process the location positioning result and order information corresponding to the current location description information to obtain text information in a specified text format. The solution output model is called to process the text information in a specified text format to obtain the boarding solution corresponding to the online car-hailing order. According to the pick-up location in the boarding solution and the current location of the driver and the passenger, the target pick-up location with the lowest pick-up cost is determined, and the pick-up solution corresponding to the target pick-up location is recommended to the driver and the passenger. The embodiment of the present application intervenes in the call process between the driver and the passenger, utilizes the integrated capabilities of multiple network models and the reinforcement learning capabilities, provides a solution for the driver and the passenger to meet smoothly, improves the communication efficiency between the driver and the passenger, reduces the probability of misunderstanding between the driver and the passenger, and promotes the completion of the order between the driver and the passenger. At the same time, recommending boarding solutions based on the way the driver and passengers arrive at the pick-up location can reduce the loss of online ride-hailing orders while ensuring the driver and passengers' taxi experience, and protect the economic benefits of drivers and platforms.

[0022] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A flowchart of the steps of a recommended method for a boarding solution provided in an embodiment of the present application;

[0024] Figure 2 A flowchart of a method for obtaining location description information provided in an embodiment of the present application;

[0025] Figure 3 A flowchart of a method for obtaining text information provided in an embodiment of the present application;

[0026] Figure 4 A flowchart of the steps of a solution output model training method provided in an embodiment of the present application;

[0027] Figure 5A flowchart of the steps of a method for obtaining a boarding solution provided in an embodiment of the present application;

[0028] Figure 6 A flowchart of a method for selecting a target pick-up location provided in an embodiment of the present application;

[0029] Figure 7 A flowchart of another method for recommending a boarding solution provided in an embodiment of the present application;

[0030] Figure 8 A schematic diagram of the structure of a recommended device for a boarding solution provided in an embodiment of the present application;

[0031] Fig. 9 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0032] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0033] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0034] Reference Figure 1 , which shows a flowchart of the steps of a recommended method for a boarding solution provided by an embodiment of the present application, such as Figure 1 As shown, the method for recommending the on-board solution may include: step 101, step 102, step 103, step 104 and step 105.

[0035] Step 101: After a passenger initiates an online car-hailing order on an online car-hailing platform, obtain voice call information between the driver and the passenger during the telephone call, as well as order information of the online car-hailing order.

[0036] In this embodiment, the online car-hailing platform is an online travel service system based on Internet technology, which connects passengers and drivers and provides a convenient and efficient way of travel. Passengers can place orders through mobile applications or websites, select the departure and destination, and the platform will send the order information to nearby drivers. After receiving the order, the driver will go to the designated location to pick up the passenger and deliver the passenger safely to the destination.

[0037] When a passenger places an online ride-hailing order on an online ride-hailing platform, and the driver accepts the order and picks up the passenger, it is possible to monitor whether there is a phone call between the driver and the passenger. If so, the voice call information between the driver and the passenger can be obtained. In actual applications, in order to protect the privacy of passengers and drivers, online ride-hailing platforms usually use virtual number technology. This means that passengers and drivers use virtual numbers provided by the platform instead of real numbers when they talk. The online ride-hailing platform can provide a call recording function, which is usually set up to resolve disputes or improve service quality. If the platform provides a call recording function and the passenger or driver agrees to the recording, then the recording information can be saved by the platform.

[0038] At the same time, the order information of the online car-hailing order can be obtained, and the order information may include but is not limited to: the current location, time, weather of both the driver and the passenger, the starting point name and longitude and latitude set for the order, the driver's navigation path, etc.

[0039] After obtaining the voice call information between the driver and the passenger during the telephone call and the order information of the online car-hailing order, execute step 102.

[0040] Step 102: calling a speech recognition model to process the voice call information to obtain current location description information of the driver and the passenger.

[0041] In this embodiment, the speech recognition model can be a model that has been adjusted in advance based on map domain knowledge information. The speech recognition model adjusted based on map domain knowledge information is a model specifically used to process speech data containing geographical location related information. The model can use specific knowledge in the map domain to improve the accuracy of speech recognition.

[0042] The implementation process of the speech recognition model adjusted based on map domain knowledge information can be as follows:

[0043] 1. Data collection and preprocessing. Collect a large amount of voice call data containing geographic location information from online ride-hailing platforms or other relevant channels. Manually annotate the collected voice data, transcribe the voice into text, and specifically annotate the geographic location information (such as street names, place names, etc.). Remove noise, silence segments, and other non-speech-related data parts to improve data quality. Incorporate specific knowledge in the map field (such as common place names, street names, landmarks, etc.) into the data in the form of a vocabulary, dictionary, or embedded vector.

[0044] 2. Model selection and architecture. Choose a suitable basic speech recognition model, such as recurrent neural network (RNN), long short-term memory network (LSTM), gated recurrent unit (GRU) or Transformer. These models perform well in processing sequence data (such as speech signals). On the basis of the basic model, add specific network layers or modules to capture map domain knowledge. For example, you can introduce an attention mechanism that includes map domain vocabulary, or use a dedicated vocabulary embedding layer to process these words.

[0045] 3. Training and optimization: Use labeled speech data to train the model. During the training process, the model will learn to map the speech signal to the corresponding text annotation, and pay special attention to the accuracy of the geographic location information. Choose a suitable loss function to measure the difference between the text output by the model and the actual annotation. For speech recognition tasks, commonly used loss functions include connectionist temporal classification (CTC) loss or sequence-to-sequence loss (such as cross entropy loss, etc.). Use optimization algorithms (such as Adam, SGD, etc.) to update the parameters of the model to minimize the loss function. During the training process, data augmentation techniques (such as adding noise, changing the speaking speed, etc.) can be used to improve the robustness of the model, and special attention should be paid to the capture and representation of map domain knowledge.

[0046] 4. Model evaluation and tuning: Use the test data set to evaluate the performance of the model. Common evaluation indicators include word error rate (WER), character error rate (CER), etc. Pay special attention to the accuracy of the model in identifying geographic location information, which can be evaluated by comparing the model output with the geographic location information in the real annotation. Tune the model based on the evaluation results, which may include adjusting the model architecture, adding training data, optimizing hyperparameters, etc.

[0047] After adjusting the speech recognition model based on map domain knowledge information, the trained model can be deployed to the online ride-hailing platform or other related applications.

[0048] After obtaining the voice call information between the driver and the passenger, the voice recognition model can be called to process the voice call information to obtain the current location description information of the driver and the passenger. Specifically, the voice recognition model can distinguish the roles in the voice call information and recognize the voice information of different roles to obtain the current location description information of the driver and the current location description information of the passenger. The current location description information may include the environmental description information of the current location of the driver and the passenger, which may include: geographic location description information, such as city name, street name, landmark building, intersection, etc. It may also include surrounding facilities, such as restaurants, shopping centers, hospitals, bus stops, subway stations, etc.

[0049] The implementation process of obtaining the current location description information of the driver and the passenger can be combined with Figure 2 This is described in detail as follows.

[0050] Reference Figure 2 , shows a flowchart of the steps of a method for obtaining location description information provided by an embodiment of the present application. Figure 2 As shown, the method for obtaining location description information may include: step 201, step 202 and step 203.

[0051] Step 201: Input the voice call information into the speech recognition model.

[0052] In this embodiment, after the voice call information between the driver and the passenger is obtained, the voice call information can be input into the speech recognition model.

[0053] Step 202: calling the speech recognition model to perform role recognition on the voice call information, and obtaining the voice information of the driver and the voice information of the passenger in the voice call information.

[0054] The speech recognition model has the ability to distinguish different speakers. Through the speech recognition model, the role of the voice call information can be identified to obtain the driver's voice information and the passenger's voice information in the voice call information.

[0055] After calling the speech recognition model to perform role recognition on the voice call information and obtaining the driver's voice information and the passenger's voice information in the voice call information, step 203 is executed.

[0056] Step 203: Recognize and process the voice information of the driver and the voice information of the passenger to obtain the demand information of the driver and the passenger, and the current location description information of the driver and the passenger.

[0057] After obtaining the voice information of the driver and the passenger in the voice call information, the voice information of the driver and the passenger can be recognized and processed to obtain the demand information of the driver and the passenger, as well as the current position description information of the driver and the passenger. The demand information can be the requirements of the driver and / or the passenger, such as the requirement that the driver's position does not move, that is, the driver's position is fixed, or the requirement of the passenger for the remaining time to reach the destination.

[0058] After the voice recognition model is called to process the voice call information and obtain the current location description information of the driver and the passenger, step 103 is executed.

[0059] Step 103: calling a text arrangement model to process the location positioning result corresponding to the current location description information and the order information to obtain text information in a specified text format.

[0060] After obtaining the current location description information of the driver and the passenger, the text sorting model can be called to process the location positioning result and order information corresponding to the current location description information to obtain text information in a specified text format. Among them, the text sorting model refers to a neural network model specifically used for text induction and sorting, which can automatically analyze and process a large amount of text data, extract key information, and display it in a specified Gernben format.

[0061] In this example, AddressCLIP technology can be used to match the current location description information of the driver and the passenger to obtain the fuzzy location matching results of the driver and the passenger, and then the text sorting model can be used to summarize the pattern location matching results and the order information to obtain text information in a specified text format. Figure 3 This is described in detail as follows.

[0062] Reference Figure 3 , shows a flowchart of the steps of a method for obtaining text information provided by an embodiment of the present application. Figure 3 As shown, the text information acquisition method may include: step 301 and step 302.

[0063] Step 301: Performing location matching processing on the current location description information of the driver and the passenger based on the AddressCLIP technology to obtain fuzzy location matching results of the driver and the passenger.

[0064] In this embodiment, AddressCLIP technology is a multimodal learning-based technology that can understand and match geographic location descriptions. AddressCLIP technology can match voice descriptions with massive address data and map domain knowledge to find the location that best matches the description.

[0065] After obtaining the current location description information of the driver and the passenger, the current location description information of the driver and the passenger can be processed for location matching based on the AddressCLIP technology to obtain the fuzzy location matching results of the driver and the passenger. In practical applications, the fuzzy location matching results of the environment described in the voice refer to the new location positioning results formed by using the AddressCLIP technology based on the environmental characteristics described by the driver and the passenger (such as "I am standing under the XXX billboard", etc.), the current location of the driver / passenger, and the massive address data (name, house number, longitude and latitude information, etc.) and map domain knowledge (property information, business information, road information, etc.) accumulated by the platform, to replace / supplement the current mobile phone positioning of the driver and the passenger, and prevent inaccurate positioning and poor signal. At the same time, the location is also calibrated based on the voice description of the driver and the passenger, and has a high degree of credibility.

[0066] After performing position matching processing on the current position description information of the driver and the passenger based on the AddressCLIP technology to obtain the fuzzy position matching result of the driver and the passenger, step 302 is executed.

[0067] Step 302: calling the text sorting model to perform text summarization and sorting processing on the fuzzy position matching result and the order information to obtain text information in the specified text format.

[0068] After performing location matching processing on the current location description information of the driver and the passenger based on AddressCLIP technology and obtaining the fuzzy location matching results of the driver and the passenger, the text sorting model can be called to perform text summarization and sorting processing on the fuzzy location matching results and the order information to obtain text information in a specified text format. The text information in the specified text format may include: the current demands of the driver and the passenger, the current mobile phone locations of the driver and the passenger, the starting point name and longitude and latitude set in the order, the driver's navigation path, the fuzzy location matching results of the current environment described in the voice, and other additional information. Other additional information may include: the emotions of the driver and the passenger, the time, the weather, etc.

[0069] After the text sorting model is called to process the location positioning result and order information corresponding to the current location description information to obtain text information in a specified text format, step 104 is executed.

[0070] Step 104: Call the solution output model to process the text information in the specified text format to obtain the boarding solution corresponding to the online car-hailing order.

[0071] The solution output model is a large language model specifically used to output solutions. It combines the latest achievements in deep learning and natural language processing technologies, and can understand and analyze large amounts of text information and generate specific and feasible solutions based on this information.

[0072] The training process of the solution output model can be combined with Figure 4 This is described in detail as follows.

[0073] Reference Figure 4 , shows a flowchart of the steps of a solution output model training method provided by an embodiment of the present application. Figure 4 As shown, the solution output model training method may include: step 401, step 402, step 403 and step 404.

[0074] Step 401: Obtain historical online ride-hailing order data, wherein the historical online ride-hailing order data includes: orders without driver-passenger call records, orders with call records, orders from drivers looking for passengers, and orders from passengers looking for drivers.

[0075] In this embodiment, when training the solution output model, historical online ride-hailing order data can be obtained. The historical online ride-hailing order data may include: orders without driver-passenger call records, orders with call records, orders from drivers looking for passengers, and orders from passengers looking for drivers, etc.

[0076] After obtaining the historical online car-hailing order data, execute step 402.

[0077] Step 402: Extract order features of the historical online car-hailing order data, where the order features include: order type features, call content features, original boarding location features, and final boarding location features.

[0078] After obtaining the historical online ride-hailing order data, the order features of the historical online ride-hailing order data can be extracted. The order features may include: order type features, call content features, original boarding location features, and final boarding location features.

[0079] After extracting the order features of the historical online ride-hailing order data, execute step 403.

[0080] Step 403: Encode the order features to obtain order encoding features.

[0081] After extracting the order features of the historical online car-hailing order data, the order features can be encoded to obtain order encoding features.

[0082] After encoding the order features to obtain the order encoding features, step 404 is executed.

[0083] Step 404: The solution output model is obtained based on the order coding feature training.

[0084] After encoding the order features to obtain the order coding features, the solution output model can be trained based on the order coding features. Specifically, the model can predict and output the best solutions for the driver and the passenger based on the input order coding features. The solution should include the actor (driver / passenger / both parties), the longitude and latitude of the new boarding location after the intervention, and the address / place name of the new boarding location. In order to train this model, a supervised learning method can be used, that is, a set of labeled solutions are provided as training data. These solutions can be statistical results based on historical data or manually labeled.

[0085] After the model training is completed, the model needs to be evaluated to check whether its performance meets the requirements. Evaluation indicators can include accuracy, recall, F1 score, etc. If the model performance is poor, you can try the following methods to optimize it:

[0086] 1. Adjust feature extraction and encoding methods.

[0087] 2. Try different deep learning algorithms or model architectures.

[0088] 3. Increase the amount of training data to improve the generalization ability of the model.

[0089] After training the solution output model, the solution output model can be used to provide boarding solutions for drivers and passengers of the current online car-hailing order. That is, the solution output model is called to process the text information in the specified text format to obtain the boarding solution corresponding to the online car-hailing order. Figure 5 This is described in detail as follows.

[0090] Reference Figure 5 , shows a flowchart of the steps of a method for obtaining a boarding solution provided by an embodiment of the present application. Figure 5 As shown, the method for obtaining a boarding solution may include: step 501 and step 502.

[0091] Step 501: Encode the text information in the specified text format to obtain text encoding features.

[0092] In this embodiment, after obtaining the text information in the specified text format, the text information in the specified text format can be encoded to obtain text encoding features. Specifically, the text information in the specified format can be encoded using methods such as bag of words model, TF-IDF, word embedding, etc., to obtain text encoding features.

[0093] Step 502: Call the solution output model to process the text encoding features to obtain the boarding solution corresponding to the online car-hailing order.

[0094] Furthermore, the solution output model can be called to process the text encoding features to obtain the boarding solution corresponding to the online car-hailing order. Specifically, after obtaining the text encoding features, they can be input into the pre-trained solution output model. This model is trained based on historical order data and corresponding solutions, and can learn how to generate the best boarding solution based on the characteristics of the current order. Among them, the boarding solution includes: the action party, the longitude and latitude information of the new boarding location, and the address information of the new boarding location. The action party is: at least one of the driver and the passenger.

[0095] In this example, the actor specifies whether the driver, the passenger, or both need to take action.

[0096] New longitude and latitude information for the boarding location: Provides a more accurate or easier-to-find boarding location to improve the encounter experience for both the driver and the passenger.

[0097] Address information of the new pickup location: In addition to the latitude and longitude, a human-readable description of the address is provided to make it easier for drivers and passengers to understand and navigate to the new location.

[0098] After calling the solution output model to process the text information in the specified text format and obtaining the boarding solution corresponding to the online car-hailing order, execute step 105.

[0099] Step 105: Determine a target pick-up location with the lowest pick-up cost among the pick-up locations based on the pick-up locations in the pick-up solution and the current locations of the driver and the passenger, and recommend a pick-up solution corresponding to the target pick-up location to the driver and the passenger.

[0100] After calling the solution output model to process the text information in the specified text format and obtaining the pick-up solution corresponding to the online car-hailing order, the pick-up cost of the pick-up location can be determined based on the pick-up location in the pick-up solution and the current locations of the driver and the passenger, and the target pick-up location with the smallest pick-up cost can be screened out from the pick-up locations, so that the pick-up solution corresponding to the target pick-up location can be recommended to the driver and the passenger.

[0101] The embodiment of the present application intervenes in the conversation process between the driver and the passenger, utilizes the ability of integrating multiple network models and the ability of strengthening learning, provides a solution for the driver and the passenger to meet smoothly, improves the efficiency of communication between the driver and the passenger, reduces the probability of misunderstanding between the driver and the passenger, and promotes the completion of the order between the driver and the passenger. At the same time, the recommendation of the boarding solution according to the way the driver and the passenger arrive at the pick-up location can reduce the loss of online car-hailing orders while ensuring the taxi experience of the driver and the passenger, and ensure the economic benefits of the driver and the platform.

[0102] In this embodiment, when recommending a boarding solution, a pre-trained reinforcement learning model can be used to perform cost evaluation, and the target pick-up location can be determined based on the cost evaluation result. Figure 6 This is described in detail as follows.

[0103] Reference Figure 6 , shows a flowchart of the steps of a target pick-up location screening method provided by an embodiment of the present application. Figure 6 As shown, the target pick-up location screening method may include: step 601, step 602 and step 603.

[0104] Step 601: Call a pre-trained reinforcement learning model to process the driver's current position and the pick-up position to obtain a first pick-up cost for the driver to arrive at the pick-up position; and process the passenger's current position and the pick-up position to obtain a second pick-up cost for the passenger to arrive at the pick-up position.

[0105] In this embodiment, after obtaining the boarding solution, a pre-trained reinforcement learning model can be called to perform cost evaluation on multiple boarding solutions for both drivers and passengers to obtain a cost evaluation result.

[0106] In practical applications, suitable reinforcement learning algorithms can be selected, such as Q-learning, deep Q network (DQN), etc., and historical data can be used to train the model so that it can learn to choose the best boarding plan under different circumstances. During the training process, different boarding plans can be tried continuously, and the strategy can be updated according to the cost of both the driver and the passenger (such as the pick-up distance, the complexity of the navigation path, time, money, comfort, etc.).

[0107] When faced with multiple new boarding solutions, the pick-up location in each boarding solution, as well as the current location of the driver and the passenger, are input into the pre-trained reinforcement learning model. The model will evaluate the cost of the pick-up location of each boarding solution based on its learned strategy and output a pick-up cost. Specifically, the reinforcement learning model can be called to process the driver's current location and the pick-up location to obtain the driver's first navigation path and first arrival time to the pick-up location. The reinforcement learning model is also called to process the passenger's current location and the pick-up location to obtain the passenger's second navigation path and second arrival time to the pick-up location. That is, the reinforcement learning model can plan the navigation path for the driver and passenger to arrive at the pick-up location based on the driver and passenger's current location and the pick-up location in the boarding solution, and predict the time it takes for the driver and passenger to reach the pick-up location.

[0108] The reinforcement learning model can then determine the first pick-up cost for the driver to arrive at the pick-up location based on the first navigation path and the first arrival time. The first pick-up cost may include time cost, fuel consumption, vehicle wear and tear, etc., and the output is a cost value. Specifically, in the driver-oriented solution, in addition to meeting the requirement of a high probability of picking up passengers, it is usually revised in accordance with the driver's current general direction of travel to avoid the driver making a large U-turn, going up and down elevated roads / tunnels, and driving too long beyond the original navigation distance. At the same time, the new boarding location will have the characteristics of being easy to observe from the driver's perspective and convenient for parking.

[0109] The reinforcement learning model determines the second pick-up cost for the passenger to arrive at the pick-up location based on the second navigation path and the second arrival time. The second pick-up cost can be time cost, walking difficulty, comfort, etc., and the output is also a cost value. In addition to meeting the requirement of a high probability of driver-passenger convergence, passenger-oriented solutions usually require visibility to passengers (i.e., easy for passengers to observe) and relatively low walking costs.

[0110] Different cost factors can be used for different roles, such as drivers and passengers. For example, passengers may pay more attention to walking distance and user willingness, while drivers may pay more attention to the overlap of driving routes, order completion probability, money, etc.

[0111] Of course, when predicting the pick-up cost, you can also input the common characteristics of the driver and the passenger, such as the emotions of both parties, the current weather, the number of drivers / passengers currently in the vicinity, etc., to ensure that the interests of both the driver and the passenger are balanced.

[0112] After obtaining the first access cost and the second access cost, step 602 is executed.

[0113] Step 602: Determine the pick-up cost corresponding to the pick-up location based on the first pick-up cost and the first weight corresponding to the driver, and the second pick-up cost and the second weight corresponding to the passenger.

[0114] After obtaining the first pick-up cost and the second pick-up cost, the pick-up cost corresponding to the pick-up location can be determined according to the first pick-up cost and the first weight corresponding to the driver, and the second pick-up cost and the second weight corresponding to the passenger. Specifically, the first pick-up cost and the first weight corresponding to the driver, and the second pick-up cost and the second weight corresponding to the passenger can be weighted summed to obtain the pick-up cost corresponding to the pick-up location, that is, the pick-up cost = the first pick-up cost × the first weight + the second pick-up cost × the second weight.

[0115] After the pick-up cost corresponding to the pick-up location is determined based on the first pick-up cost and the first weight corresponding to the driver, and the second pick-up cost and the second weight corresponding to the passenger, step 603 is executed.

[0116] Step 603: Filter out the target pick-up location with the minimum pick-up cost from the pick-up locations.

[0117] Furthermore, the target pick-up location with the lowest pick-up cost can be screened out from the pick-up locations.

[0118] By comprehensively considering the costs of both drivers and passengers (such as time, money, comfort, etc.), the model of this application can find a solution that balances the interests of both parties, which helps to optimize the allocation of traffic resources, reduce unnecessary waste, and improve overall efficiency. At the same time, the solution with the lowest cost often means higher passenger satisfaction and better driver benefits. Recommending such a solution can improve users' overall evaluation of travel services and enhance user stickiness.

[0119] In this embodiment, when recommending the on-board solution, a navigation planning path and text guidance information can be generated according to the on-board solution to make the recommendation at the same time. Figure 7 This is described in detail as follows.

[0120] Reference Figure 7 , shows a flowchart of another method for recommending a boarding solution provided by an embodiment of the present application. Figure 7 As shown, the on-board solution recommendation method may include: step 701 and step 702.

[0121] Step 701: Generate a navigation planning path and text guidance information for the driver and / or the passenger according to the boarding solution corresponding to the target pick-up location.

[0122] In this embodiment, after the target pick-up location is obtained, the navigation planning path and text guidance information of the driver and / or passenger can be generated according to the boarding solution corresponding to the target pick-up location. Specifically, the boarding solution corresponding to the target pick-up location includes a new boarding location point, and the navigation path can be planned in combination with the new boarding location point and the current location information of the driver / or passenger. At the same time, text guidance information can be generated to guide the driver and / or passenger to the new boarding location point.

[0123] Step 702: Recommending the boarding solution corresponding to the target boarding location, the navigation planning path and the text guidance information to the driver and / or the passenger.

[0124] Furthermore, the boarding solution, navigation planning path and text guidance information corresponding to the target boarding location can be recommended to the driver and / or passenger.

[0125] In this embodiment, the recommendation method for boarding solutions for drivers and passengers can be differentiated.

[0126] For drivers, they can directly adjust the end point of the driver's current pick-up navigation (i.e. replace it with a new pick-up point) on the driver-side APP, and plan a new navigation route based on the driver's current location and the new pick-up point.

[0127] For passengers, a prompt can be generated on the passenger-side APP: inform the passenger that there is a new recommended plan for their original starting point, and let the passenger know and click the button, and then push it to the passenger. If the passenger does not click the button, the passenger's original starting point will not be changed. If the passenger accepts to try the new boarding location, the new boarding location can be displayed on the passenger's APP, and a walking guide containing pictures and text can be generated. The guidance can also include path planning and landmark picture display.

[0128] The embodiment of the present application achieves the effect of guiding the driver and passenger to meet by displaying the new boarding point location, navigation planning route and text guidance on the driver side and the passenger side respectively according to the solution content.

[0129] The method for recommending a boarding solution provided in the embodiment of the present application obtains the voice call information of the driver and the passenger during the telephone call and the order information of the online car-hailing order after the passenger initiates the online car-hailing order on the online car-hailing platform. The pre-trained speech recognition model adjusted based on map domain knowledge information is called to process the voice call information to obtain the current location description information of the driver and the passenger. The text sorting model is called to process the location positioning result and order information corresponding to the current location description information to obtain text information in a specified text format. The solution output model is called to process the text information in a specified text format to obtain the boarding solution corresponding to the online car-hailing order. According to the pick-up location in the boarding solution, the current location of the driver and the passenger, the target pick-up location with the lowest pick-up cost is determined, and the boarding solution corresponding to the target pick-up location is recommended to the driver and the passenger. The embodiment of the present application intervenes in the call process between the driver and the passenger, utilizes the integrated capabilities of multiple network models and the reinforcement learning capabilities, provides a solution for the driver and the passenger to meet smoothly, improves the communication efficiency between the driver and the passenger, reduces the probability of misunderstanding between the driver and the passenger, and promotes the completion of the order between the driver and the passenger. At the same time, recommending boarding solutions based on the way the driver and passengers arrive at the pick-up location can reduce the loss of online ride-hailing orders while ensuring the driver and passengers' taxi experience, and protect the economic benefits of drivers and platforms.

[0130] Reference Figure 8 , shows a schematic diagram of the structure of a recommended device for a boarding solution provided by an embodiment of the present application, such as Figure 8 As shown, the vehicle solution recommendation device 800 may include the following modules:

[0131] The voice call information acquisition module 810 is used to acquire the voice call information of the phone call between the driver and the passenger, as well as the order information of the online car-hailing order, after the passenger initiates the online car-hailing order on the online car-hailing platform;

[0132] A location description information acquisition module 820 is used to call a speech recognition model to process the voice call information to obtain current location description information of the driver and the passenger;

[0133] The text information acquisition module 830 is used to call the text arrangement model to process the location positioning result corresponding to the current location description information and the order information to obtain text information in a specified text format;

[0134] The boarding solution acquisition module 840 is used to call the solution output model to process the text information in the specified text format to obtain the boarding solution corresponding to the online car-hailing order;

[0135] The boarding solution recommendation module 850 is used to determine the target pick-up location with the lowest pick-up cost among the pick-up locations based on the pick-up location in the boarding solution and the current locations of the driver and the passenger, and recommend the boarding solution corresponding to the target pick-up location to the driver and the passenger.

[0136] Optionally, the location description information acquisition module includes:

[0137] A voice call information input unit, used to input the voice call information into the speech recognition model;

[0138] A voice information recognition unit, used to call the voice recognition model to perform role recognition on the voice call information, and obtain the voice information of the driver and the voice information of the passenger in the voice call information;

[0139] The location description information acquisition unit is used to recognize and process the voice information of the driver and the voice information of the passenger to obtain the demand information of the driver and the passenger, and the current location description information of the driver and the passenger.

[0140] Optionally, the text information acquisition module includes:

[0141] A fuzzy matching result acquisition unit, used for performing position matching processing on the current position description information of the driver and the passenger based on AddressCLIP technology to obtain fuzzy position matching results of the driver and the passenger;

[0142] A text information acquisition unit, used for calling the text arrangement model to perform text summarization and arrangement processing on the fuzzy position matching result and the order information to obtain text information in the specified text format;

[0143] Among them, the text information in the specified format includes: the demand information of the driver and the passenger, the location information of the driver and the passenger, the starting point name and starting longitude and latitude of the online car-hailing order, the driver's navigation path information and the fuzzy location matching result.

[0144] Optionally, the device further comprises:

[0145] A historical order data acquisition module is used to acquire historical online car-hailing order data, wherein the historical online car-hailing order data includes: orders without driver-passenger call records, orders with call records, orders from drivers looking for passengers, and orders from passengers looking for drivers;

[0146] An order feature extraction module is used to extract order features of the historical online car-hailing order data, wherein the order features include: order type features, call content features, original boarding location features, and final boarding location features;

[0147] An order coding feature acquisition module is used to encode the order features to obtain order coding features;

[0148] A model training module is used to obtain the solution output model based on the order coding feature training.

[0149] Optionally, the on-board solution acquisition module includes:

[0150] A text encoding feature acquisition module is used to encode the text information in the specified text format to obtain text encoding features;

[0151] A boarding solution acquisition unit, used to call the solution output model to process the text encoding features to obtain a boarding solution corresponding to the online car-hailing order;

[0152] The boarding solution includes: an action party, longitude and latitude information of a new boarding location, and address information of the new boarding location, and the action party is: at least one of the driver and the passenger.

[0153] Optionally, the on-board solution recommendation module includes:

[0154] a pick-up cost acquisition unit, configured to call a pre-trained reinforcement learning model to process the current position of the driver and the pick-up position to obtain a first pick-up cost for the driver to arrive at the pick-up position; and to process the current position of the passenger and the pick-up position to obtain a second pick-up cost for the passenger to arrive at the pick-up position;

[0155] a pick-up cost determining unit, configured to determine a pick-up cost corresponding to the pick-up location according to the first pick-up cost and a first weight corresponding to the driver, and the second pick-up cost and a second weight corresponding to the passenger;

[0156] The target pick-up location screening unit is used to screen out the target pick-up location with the minimum pick-up cost from the pick-up locations.

[0157] Optionally, the pick-up cost acquisition unit includes:

[0158] a navigation time acquisition subunit, configured to call the reinforcement learning model to process the current position of the driver and the pick-up position, and obtain a first navigation path and a first arrival time for the driver to arrive at the pick-up position; and to process the current position of the passenger and the pick-up position, and obtain a second navigation path and a second arrival time for the passenger to arrive at the pick-up position;

[0159] A first pick-up cost determination subunit, configured to determine a first pick-up cost for the driver to arrive at the pick-up location based on the first navigation route and the first arrival time;

[0160] The second pick-up cost determination subunit is used to determine a second pick-up cost for the driver to arrive at the pick-up location based on the second navigation route and the second arrival time.

[0161] Optionally, the on-board solution recommendation module includes:

[0162] A navigation planning path generation unit, configured to generate a navigation planning path and text guidance information for the driver and / or the passenger according to a boarding solution corresponding to the target pick-up location;

[0163] The second boarding solution recommendation unit is used to recommend the boarding solution corresponding to the target pick-up location, the navigation planning path and the text guidance information to the driver and / or the passenger.

[0164] The device for recommending the boarding solution provided in the embodiment of the present application obtains the voice call information of the driver and the passenger during the telephone call and the order information of the online car-hailing order after the passenger initiates the online car-hailing order on the online car-hailing platform. The pre-trained speech recognition model adjusted based on map domain knowledge information is called to process the voice call information to obtain the current location description information of the driver and the passenger. The text sorting model is called to process the location positioning result and order information corresponding to the current location description information to obtain text information in a specified text format. The solution output model is called to process the text information in a specified text format to obtain the boarding solution corresponding to the online car-hailing order. According to the pick-up location in the boarding solution, the current location of the driver and the passenger, the target pick-up location with the lowest pick-up cost is determined, and the boarding solution corresponding to the target pick-up location is recommended to the driver and the passenger. The embodiment of the present application intervenes in the call process between the driver and the passenger, utilizes the integrated capabilities of multiple network models and the reinforcement learning capabilities, provides a solution for the driver and the passenger to meet smoothly, improves the communication efficiency between the driver and the passenger, reduces the probability of misunderstanding between the driver and the passenger, and promotes the completion of the order between the driver and the passenger. At the same time, recommending boarding solutions based on the way the driver and passengers arrive at the pick-up location can reduce the loss of online ride-hailing orders while ensuring the driver and passengers' taxi experience, and protect the economic benefits of drivers and platforms.

[0165] An embodiment of the present application also provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the above-mentioned method for recommending a boarding solution when executed by the processor.

[0166] Fig. 9 FIG. 1 is a schematic diagram showing the structure of an electronic device 900 according to an embodiment of the present invention. Fig. 9 As shown, the electronic device 900 includes a central processing unit (CPU) 901, which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) 902 or computer program instructions loaded from a storage unit 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the electronic device 900 can also be stored. The CPU 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0167] Multiple components in the electronic device 900 are connected to the I / O interface 905, including: an input unit 906, such as a keyboard, a mouse, a microphone, etc.; an output unit 907, such as various types of displays, speakers, etc.; a storage unit 908, such as a disk, an optical disk, etc.; and a communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the electronic device 900 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0168] The various processes and processing described above may be performed by the processing unit 901. For example, the method of any of the above embodiments may be implemented as a computer software program, which is tangibly contained in a computer-readable medium, such as the storage unit 908. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 900 via the ROM 902 and / or the communication unit 909. When the computer program is loaded into the RAM 903 and executed by the CPU 901, one or more actions in the method described above may be performed.

[0169] In addition, an embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the recommended method for the above-mentioned boarding solution.

[0170] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0171] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, devices, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0172] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, terminals (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0173] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0174] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal so that a series of operating steps are executed on the computer or other programmable terminal to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable terminal. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0175] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present application.

[0176] Finally, it should be noted that, in this article, relational terms such as first and second, etc. 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 terminal 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 terminal. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or terminal including the elements.

[0177] The above is a detailed introduction to a recommended method for a boarding solution, a recommended device for a boarding solution, an electronic device and a computer-readable storage medium provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for recommending a vehicle solution, characterized in that: The method comprises: After the passenger places an online car-hailing order on the online car-hailing platform, obtain the voice call information between the driver and the passenger during the telephone call, as well as the order information of the online car-hailing order; Invoking a speech recognition model to process the voice call information to obtain current location description information of the driver and the passenger; Calling a text arrangement model to process the location positioning result corresponding to the current location description information and the order information to obtain text information in a specified text format; Calling the solution output model to process the text information in the specified text format to obtain a boarding solution corresponding to the online car-hailing order; According to the pick-up location in the boarding solution and the current locations of the driver and the passenger, a target pick-up location with the minimum pick-up cost among the pick-up locations is determined, and a boarding solution corresponding to the target pick-up location is recommended to the driver and the passenger.

2. The method according to claim 1, characterized in that: The calling of the speech recognition model to process the voice call information to obtain the current location description information of the driver and the passenger includes: inputting the voice call information into the voice recognition model; Calling the speech recognition model to perform role recognition on the voice call information, and obtaining the voice information of the driver and the voice information of the passenger in the voice call information; The voice information of the driver and the voice information of the passenger are recognized and processed to obtain the demand information of the driver and the passenger, as well as the current position description information of the driver and the passenger.

3. The method according to claim 1, characterized in that The calling text arrangement model processes the location positioning result corresponding to the current location description information and the order information to obtain text information in a specified text format, including: Performing location matching processing on the current location description information of the driver and the passenger based on AddressCLIP technology to obtain fuzzy location matching results of the driver and the passenger; Calling the text sorting model to perform text summarization and sorting processing on the fuzzy position matching result and the order information to obtain text information in the specified text format; Among them, the text information in the specified format includes: the demand information of the driver and the passenger, the location information of the driver and the passenger, the starting point name and starting longitude and latitude of the online car-hailing order, the driver's navigation path information and the fuzzy location matching result.

4. The method according to claim 1, characterized in that Before the calling solution output model to process the text information in the specified text format to obtain the boarding solution corresponding to the online car-hailing order, the method further includes: Obtaining historical online ride-hailing order data, the historical online ride-hailing order data including: orders without driver-passenger call records, orders with call records, orders from drivers looking for passengers, and orders from passengers looking for drivers; Extracting order features of the historical online car-hailing order data, the order features including: order type features, call content features, original boarding location features, and final boarding location features; Encoding the order features to obtain order encoding features; The solution output model is obtained by training based on the order coding features.

5. The method according to claim 1, characterized in that The calling solution output model processes the text information in the specified text format to obtain a boarding solution corresponding to the online car-hailing order, including: Encoding the text information in the specified text format to obtain text encoding features; Calling the solution output model to process the text encoding features to obtain a boarding solution corresponding to the online car-hailing order; The boarding solution includes: an action party, longitude and latitude information of a new boarding location, and address information of the new boarding location, and the action party is: at least one of the driver and the passenger.

6. The method according to claim 1, characterized in that The step of determining a target pick-up location with the minimum pick-up cost among the pick-up locations according to the pick-up location in the boarding solution and the current locations of the driver and the passenger includes: Invoke a pre-trained reinforcement learning model to process the current position of the driver and the pick-up position to obtain a first pick-up cost for the driver to reach the pick-up position; and process the current position of the passenger and the pick-up position to obtain a second pick-up cost for the passenger to reach the pick-up position; Determine the pick-up price corresponding to the pick-up location according to the first pick-up price and the first weight corresponding to the driver, and the second pick-up price and the second weight corresponding to the passenger; The target pick-up location with the minimum pick-up cost is selected from the pick-up locations.

7. The method according to claim 6, characterized in that The calling of the pre-trained reinforcement learning model to process the current position of the driver and the pick-up position to obtain a first pick-up cost for the driver to arrive at the pick-up position; and the calling of the pre-trained reinforcement learning model to process the current position of the passenger and the pick-up position to obtain a second pick-up cost for the passenger to arrive at the pick-up position, includes: The reinforcement learning model is called to process the current position of the driver and the pick-up position to obtain a first navigation path and a first arrival time of the driver arriving at the pick-up position; and the current position of the passenger and the pick-up position are processed to obtain a second navigation path and a second arrival time of the passenger arriving at the pick-up position; Determining a first pick-up cost for the driver to arrive at the pick-up location based on the first navigation route and the first arrival time; A second pick-up cost for the driver to arrive at the pick-up location is determined based on the second navigation route and the second arrival time.

8. The method according to claim 1, characterized in that The recommending to the driver and the passenger a boarding solution corresponding to the target pick-up location includes: Generate a navigation planning path and text guidance information for the driver and / or the passenger according to the boarding solution corresponding to the target pick-up location; The boarding solution corresponding to the target pick-up location, the navigation planning path and the text guidance information are recommended to the driver and / or the passenger.

9. A device for recommending a vehicle solution, characterized in that: The device comprises: A voice call information acquisition module is used to obtain the voice call information of the driver and the passenger during the telephone call after the passenger initiates an online car-hailing order on the online car-hailing platform, as well as the order information of the online car-hailing order; A location description information acquisition module, used to call a speech recognition model to process the voice call information and obtain current location description information of the driver and the passenger; A text information acquisition module, used for invoking a text arrangement model to process the position positioning result corresponding to the current position description information and the order information to obtain text information in a specified text format; A boarding solution acquisition module, used to call the solution output model to process the text information in the specified text format to obtain the boarding solution corresponding to the online car-hailing order; The boarding solution recommendation module is used to determine the target pick-up location with the lowest pick-up cost among the pick-up locations based on the pick-up location in the boarding solution and the current locations of the driver and the passenger, and recommend the boarding solution corresponding to the target pick-up location to the driver and the passenger.

10. An electronic device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for recommending a boarding solution according to any one of claims 1 to 8 when executing the program.

11. A computer-readable storage medium, characterized in that: When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method for recommending a vehicle solution according to any one of claims 1 to 8.

Images