Recommendation method and device, electronic equipment, storage medium and computer program product

Abstract

The invention relates to a recommendation method and device, electronic equipment, a storage medium and a computer program product, and is applied to a trained recommendation model, and the recommendation model comprises a first acquisition module, a first gating loop unit GRU module and a first interest learning module. The method comprises the following steps: acquiring vehicle environment data in a vehicle running process, behavior data of a user of the vehicle and preference data of the user for vehicle function service by utilizing a first acquisition module; using the first GRU module to determine a vehicle operation scene based on the behavior data and the vehicle environment data, and generating an initial vehicle function service based on the vehicle operation scene; and adjusting the initial vehicle function service based on the preference data by using the first interest learning module to obtain a target vehicle function service, and recommending a function service of a vehicle to the user based on the target vehicle function service.

Description

Technical Field

[0001] The present disclosure relates to the field of vehicle technology, and in particular to a recommendation method, device, electronic device, storage medium, and computer program product. Background Art

[0002] As vehicles become increasingly intelligent, their functional services are becoming increasingly diverse, significantly enhancing the user's travel experience. However, with this proliferation of in-vehicle services, users often face the challenges of cumbersome user interfaces and complex function switching. This is especially true while driving, where users need to quickly locate the required service across multiple functions. However, these tasks are often difficult to achieve due to cumbersome operation steps or unintuitive interface prompts. Therefore, intelligently recommending vehicle functional services to users without requiring them to search for each service themselves is crucial. However, current methods for recommending vehicle functional services suffer from poor accuracy and flexibility. Summary of the Invention

[0003] To overcome the problems existing in the related art, the present disclosure provides a recommendation method, apparatus, electronic device, storage medium and computer program product.

[0004] According to a first aspect of an embodiment of the present disclosure, a recommendation method is provided, which is applied to a trained recommendation model, wherein the recommendation model includes a first acquisition module, a first gated recurrent unit (GRU) module, and a first interest learning module. The method includes:

[0005] Using the first acquisition module to acquire vehicle environment data during vehicle operation, behavior data of a user of the vehicle, and preference data of the user for vehicle function services;

[0006] Determine, using the first GRU module, a vehicle operation scenario based on the behavior data and the vehicle environment data, and generate an initial vehicle function service based on the vehicle operation scenario;

[0007] The first interest learning module is used to adjust the initial vehicle function service based on the preference data to obtain a target vehicle function service, and the vehicle function service is recommended to the user based on the target vehicle function service.

[0008] In some embodiments, the training method of the recommendation model includes:

[0009] A sample data set is acquired using a second acquisition module of a preset neural network model; wherein the sample data set includes a plurality of time series samples, and each time series sample includes sample vehicle environment data, sample user behavior data, sample user preference data for vehicle function services, and sample vehicle function services;

[0010] Using the second GRU module of the neural network model, for each time series sample, a correlation relationship is constructed between the sample vehicle environment data and the sample user's behavior data to obtain a sample vehicle operation scenario, and a prediction is performed based on the sample vehicle operation scenario to obtain a sample initial vehicle function service;

[0011] Using the second interest learning module of the neural network model, for each time series sample, based on the sample user's preference data for vehicle function services, adjust the sample initial vehicle function services corresponding to the time series sample to obtain a vehicle function service prediction result for the time series sample;

[0012] Based on the difference between the vehicle function service prediction result of each time series sample and the sample vehicle function service, the parameters of the neural network model are adjusted to obtain the recommendation model.

[0013] In some embodiments, the step of establishing an association between the sample vehicle environment data and the sample user behavior data to obtain a sample vehicle operation scenario includes:

[0014] In the case where the sample user's behavior data includes multiple items, an association relationship between any two pieces of behavior data and an association relationship between each piece of behavior data and the sample vehicle environment data are constructed to obtain the sample vehicle operation scenario.

[0015] In some embodiments, the step of acquiring the sample data set using the second acquisition module of the preset neural network model includes:

[0016] Utilizing the second acquisition module to acquire a plurality of historical sample vehicle environment data, a plurality of historical sample user behavior data, a plurality of sample user historical preference data for vehicle function services, and a plurality of historical sample vehicle function services;

[0017] Based on preset data processing rules, data segmentation processing is performed on the multiple historical vehicle environment data samples, the multiple historical sample user behavior data, the multiple sample users' historical preference data for vehicle function services, and the multiple historical vehicle function service samples to obtain multiple historical time series samples, and the multiple historical time series samples are used as the sample data set; wherein the data processing rules are used to indicate rules for segmentation into time series samples, and the data processing rules include at least one of the following:

[0018] Classification based on vehicle journey; classification based on the driving time of a vehicle in a trip; classification based on user identity combined with the driving time of a vehicle in a trip.

[0019] In some embodiments, the method for determining the user's preference data for vehicle function services includes at least one of the following:

[0020] Determined based on the user's identity information data;

[0021] Determined based on the user's historical interaction data with vehicle function services.

[0022] In some embodiments, the target vehicle functional services obtained by the first interest learning module include multiple ones, the recommendation model further includes an intention reasoning module, and the method further includes:

[0023] The intention reasoning module is used to sort the multiple target functional services based on preset intervention rules, and a functional service recommended to the user is determined from the sorted multiple target functional services.

[0024] In some embodiments, the method further comprises:

[0025] Obtaining feedback from the user on the recommended functional service;

[0026] Parameters of the recommendation model are adjusted based on the vehicle environment data, the behavior data of the user of the vehicle, the preference data of the user for vehicle function services, and the feedback.

[0027] According to a second aspect of an embodiment of the present disclosure, a recommendation device is provided, which is applied to a trained recommendation model, wherein the recommendation model includes a first acquisition module, a first gated recurrent unit (GRU) module, and a first interest learning module. The device includes:

[0028] a third acquisition module configured to use the first acquisition module to acquire vehicle environment data during vehicle operation, behavior data of a user of the vehicle, and preference data of the user for vehicle function services;

[0029] a generation module configured to, using the first GRU module, determine a vehicle operation scenario based on the behavior data and the vehicle environment data, and generate an initial vehicle function service based on the vehicle operation scenario;

[0030] The recommendation module is configured to use the first interest learning module to adjust the initial vehicle function service based on the preference data to obtain the target vehicle function service to be recommended, and recommend vehicle function services to the user based on the target vehicle function service.

[0031] In some embodiments, the apparatus further comprises:

[0032] The recommendation model training module is configured to obtain a sample data set using the second acquisition module of a preset neural network model; wherein the sample data set includes multiple time series samples, and a time series sample includes sample vehicle environment data, sample user behavior data, sample user preference data for vehicle function services, and sample vehicle function services; using the second GRU module of the neural network model to construct an association relationship between the sample vehicle environment data and the sample user behavior data for each time series sample, obtain a sample vehicle operation scenario, and make a prediction based on the sample vehicle operation scenario to obtain a sample initial vehicle function service; using the second interest learning module of the neural network model to adjust the sample initial vehicle function service corresponding to the time series sample based on the sample user preference data for vehicle function services for each time series sample, to obtain a vehicle function service prediction result of the time series sample; based on the difference between the vehicle function service prediction result of each time series sample and the sample vehicle function service, adjust the parameters of the neural network model to obtain the recommendation model.

[0033] In some embodiments, the recommendation model training module is further configured to construct an association relationship between any two behavior data and an association relationship between each behavior data and the sample vehicle environment data when the sample user's behavior data includes multiple items, so as to obtain the sample vehicle operation scenario.

[0034] In some embodiments, the recommendation model training module is further configured to use the second acquisition module to obtain multiple historical sample vehicle environment data, multiple historical sample user behavior data, multiple sample users' historical preference data for vehicle function services, and multiple historical sample vehicle function services; based on preset data processing rules, the multiple historical vehicle environment data samples, the multiple historical sample user behavior data, the multiple sample users' historical preference data for vehicle function services, and the multiple historical vehicle function service samples are subjected to data division processing to obtain multiple historical time series samples, and the multiple historical time series samples are used as the sample data set; wherein, the data processing rules are used to indicate the rules for division into time series samples, and the data processing rules include at least one of the following: division based on vehicle trip; division based on the driving time of the vehicle in a trip; division based on user identity combined with the driving time of the vehicle in a trip.

[0035] In some embodiments, the apparatus further comprises:

[0036] The preference data determination module is configured to determine the user's preference data for vehicle function services in a method including at least one of the following: determining based on the user's identity information data; determining based on the user's historical interaction data with the vehicle function services.

[0037] In some embodiments, the target vehicle functional services obtained by the first interest learning module include multiple ones, the recommendation model further includes an intention reasoning module, and the device further includes:

[0038] The function service determination module is configured to use the intention reasoning module to sort the multiple target function services based on preset intervention rules, and determine the function service recommended to the user from the sorted multiple target function services.

[0039] In some embodiments, the apparatus further comprises:

[0040] a fourth acquisition module, configured to obtain feedback from the user on the recommended functional service;

[0041] An adjustment module is configured to adjust the parameters of the recommendation model based on the vehicle environment data, the behavior data of the user of the vehicle, the preference data of the user for vehicle function services and the feedback.

[0042] According to a third aspect of an embodiment of the present disclosure, there is provided an electronic device, including:

[0043] processor;

[0044] memory for storing computer programs or instructions;

[0045] The processor executes the computer program or instructions to implement the steps of the recommendation method described in the first aspect above.

[0046] According to a fourth aspect of an embodiment of the present disclosure, a non-temporary computer-readable storage medium is provided, which stores a computer program or instructions. When the computer program or instructions in the storage medium are executed by a processor, the steps of the recommended method described in the first aspect above are implemented.

[0047] According to a fifth aspect of an embodiment of the present disclosure, a computer program product is provided, comprising a computer program or instructions, which, when executed by a processor, implement the steps of the recommendation method described in the first aspect above.

[0048] The technical solutions provided by the embodiments of the present disclosure may have the following beneficial effects:

[0049] In an embodiment of the present disclosure, the electronic device uses vehicle environment data during vehicle operation, vehicle user behavior data, and user preference data for vehicle function services acquired by a first acquisition module in a trained recommendation model, and then uses the first GRU module in the trained recommendation model to determine a vehicle operation scenario based on the behavior data and vehicle environment data, and generates an initial vehicle function service based on the vehicle operation scenario. The electronic device also uses the first interest learning module in the trained recommendation model to adjust the initial vehicle function service based on the preference data to obtain a target vehicle function service, and recommends vehicle function services to the user based on the target vehicle function service. On the one hand, the recommendation model is a trained model with high accuracy. On the other hand, the embodiment of the present disclosure determines the vehicle operation scenario based on the user behavior data and vehicle environment data during vehicle operation, which can improve the accuracy and uniqueness of the determined vehicle operation scenario, thereby improving the accuracy of the initial vehicle function service generated based on the weighing operation scenario. On the other hand, the embodiment of the present disclosure adjusts the initial vehicle function service based on the user preference data for vehicle function services to obtain the target function service, and then recommends vehicle function services to the user based on the target vehicle function service. This can make the recommended function services more in line with the user's personalized needs, improve user experience and satisfaction, and have high intelligence.

[0050] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0052] Figure 1 The figure is a flowchart showing a recommendation method according to an exemplary embodiment.

[0053] Figure 2 The figure is a flowchart illustrating an example of a recommendation method according to an exemplary embodiment.

[0054] Figure 3 The figure is an architectural diagram of a recommendation model according to an exemplary embodiment.

[0055] Figure 4 The figure is a block diagram of a recommendation device according to an exemplary embodiment.

[0056] Figure 5 The figure is a structural block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0057] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.

[0058] Figure 1 FIG. 1 is a flow chart showing a recommendation method according to an exemplary embodiment. Figure 1 As shown, the method mainly includes the following steps:

[0059] S11, using the first acquisition module to acquire vehicle environment data during vehicle operation, behavior data of the vehicle user, and preference data of the user for vehicle function services;

[0060] S12. Determine, using the first GRU module, a vehicle operation scenario based on the behavior data and the vehicle environment data, and generate an initial vehicle function service based on the vehicle operation scenario;

[0061] S13. Utilize the first interest learning module to adjust the initial vehicle function service based on the preference data to obtain a target vehicle function service, and recommend vehicle function services to the user based on the target vehicle function service.

[0062] The recommendation method provided in the embodiments of the present disclosure may be executed by terminal devices such as user equipment (UE), mobile devices, user terminals, mobile phones, tablet computers, personal digital assistants (PDAs), handheld devices, computing devices, vehicle-mounted devices, wearable devices, etc., which have built-in trained recommendation models; or by cloud devices such as cloud servers, which have built-in trained recommendation models. The embodiments of the present disclosure do not limit the execution subjects. For ease of description, the embodiments of the present disclosure are described with electronic devices as the execution subjects. It should be noted that the built-in trained recommendation model of the electronic device includes a first acquisition module, a first gated recurrent unit GRU module, and a first interest learning module. The electronic device executes based on the built-in trained recommendation model. Figure 1 The training methods in .

[0063] In step S11, the electronic device uses a first acquisition module to acquire vehicle environment data during vehicle operation, vehicle user behavior data, and user preference data for vehicle functional services. The vehicle environment data includes vehicle internal data and / or vehicle external environment data, and the vehicle user behavior data includes data generated by the user's interaction with the vehicle and / or data generated by the user's own actions.

[0064] In the disclosed embodiments, vehicle internal data includes vehicle speed, vehicle temperature, status of various vehicle functions and services, vehicle engine status, vehicle tire pressure, and remaining fuel. Electronic devices can obtain this data using the vehicle's built-in sensors, the vehicle control system, or third-party interfaces. For example, electronic devices can use tire pressure sensors to monitor real-time tire pressure during vehicle operation; and can use fuel consumption monitoring systems to monitor dynamic remaining fuel levels during vehicle operation. It should be noted that vehicle internal data includes any data reflecting the vehicle's physical state, the operating status of various vehicle systems, and the status of functional services during vehicle operation, and is not limited to the aforementioned exemplary data. Vehicle internal data also includes data indicating the vehicle's responses to user interactions during operation. For example, this data may include the indications of various indicator lights, the reversing camera display and control reminders, the invocation of the dashcam, navigation information displayed on the central control screen, vehicle status, entertainment content, navigation instructions provided by the in-vehicle voice assistant, and information queries.

[0065] In the disclosed embodiment, the vehicle's external environment data includes weather, wind level, humidity, date, vehicle operation time period (such as whether it is rush hour), location, road conditions, congestion, traffic control conditions, etc., and the electronic device can obtain it using the vehicle's built-in sensors, vehicle control system or third-party interface. For example, the electronic device can monitor the traffic congestion and road obstacles on the road ahead through road sensors and high-precision lidar. The electronic device can also obtain the location-based service (LBS) trajectory through a third-party interface. The LBS trajectory is the movement trajectory data obtained through the user's navigation, location sharing, etc. during the vehicle operation through positioning systems, base station positioning or WIFI positioning technologies. These data include the user's latitude and longitude, timestamp, speed, etc. It should be noted that the vehicle's external environment data includes any data that can reflect the vehicle's external natural environment, road environment, traffic environment, etc. during the vehicle's operation, and is not limited to the above-mentioned exemplary data.

[0066] In the embodiments of the present disclosure, data generated by user-vehicle interactions may include data generated by user interactions with functional services within the vehicle during vehicle operation. For example, this may include a user enabling interior defogger, turning on music or a radio station, or using the vehicle's navigation system to mark waypoints along a travel route. It should be noted that data generated by user-vehicle interactions include any data set representing interactions between the vehicle and the user, and is not limited to the aforementioned exemplary data.

[0067] In the embodiments of the present disclosure, the data generated by the user's own actions includes various behaviors performed by the user without interacting with the vehicle. For example, the various behaviors performed by the user without interacting with the vehicle may include sleeping, putting on makeup, reading, etc., and can be captured by monitoring equipment such as the vehicle's built-in camera. The user can be the user driving the vehicle or the user riding in the vehicle, and this embodiment of the present disclosure does not limit this.

[0068] In the embodiment of the present disclosure, the user's preference data for vehicle function services may be data stored in a memory or recommendation model built into the electronic device. It should be noted that the user's preference data for vehicle function services may include the user's preference data for vehicle function services in different vehicle operation scenarios. For example, the user's preference data for vehicle function services on rainy days may be different from the user's preference data for vehicle function services on sunny days. For example, the user usually likes to adjust the vehicle air-conditioning temperature to 26 degrees on rainy days, and usually likes to adjust the vehicle air-conditioning temperature to 23 degrees on sunny days.

[0069] In the embodiment of the present disclosure, the method for determining the user preference data for vehicle function services includes at least one of the following:

[0070] Determined based on the user's identity information data;

[0071] Determined based on the user's historical interaction data with vehicle function services.

[0072] In an embodiment of the present disclosure, when a user is driving a vehicle for the first time or riding in a vehicle for the first time, and there is no historical interaction data between the user and vehicle function services in the electronic device or recommendation model, the electronic device may determine the user's preference data for vehicle function services based on the user's identity information data, wherein the user's identity information may include data such as the user's age, gender, occupation, business tags, etc. The electronic device may determine users with similar identities to the user based on the user's identity information, and determine the preference data of users with similar identities to the user for vehicle function services based on the historical behavior data of users with similar identities to the user, and use the preference data of users with similar identities to the user for vehicle function services as the user's preference data for vehicle function services.

[0073] In an embodiment of the present disclosure, when a user is not driving a vehicle for the first time or riding in a vehicle for the first time, and the electronic device or recommendation model has the user's historical interaction data with vehicle function services, the user's preference data for vehicle function services can be determined based on the user's historical interaction data with vehicle function services. For example, the electronic device can analyze the user's operation records of air conditioning temperature adjustment, navigation preferences, entertainment content selection and other functions when riding or driving a vehicle multiple times in the past, and count the user's preference trends in different scenarios, such as preferring to set a lower air conditioning temperature in hot weather, or preferring to choose a specific type of music during long-distance driving, or ordering coffee when turning on navigation, etc. It should be noted that the electronic device can apply clustering algorithms (such as K-means clustering algorithms) or classification algorithms (such as decision trees) to analyze the user's historical interaction data with vehicle function services to identify and mine the user's preference data for vehicle function services, and can also sort the multiple preference data when the user's preference data for vehicle function services includes multiple data.

[0074] In an embodiment of the present disclosure, the electronic device may utilize the first acquisition module to regularly acquire vehicle environment data during vehicle operation, vehicle user behavior data, and user preference data for vehicle functional services. For example, the electronic device may utilize the first acquisition module to acquire vehicle environment data during vehicle operation, vehicle user behavior data, and user preference data for vehicle functional services at preset time intervals. The preset time may be a set value, a default value of a recommendation model, or a value preset by the user, such as 5 minutes or 10 minutes, etc., and the embodiment of the present disclosure does not impose any restrictions on this.

[0075] In step S12, the electronic device uses the first GRU module to determine the vehicle operation scenario based on the behavioral data and vehicle environment data, and generates initial vehicle function services based on the vehicle operation scenario. The GRU module is a special recurrent neural network (RNN) unit that solves the gradient vanishing and gradient exploding problems in traditional RNNs by introducing a "gate" mechanism, thereby more effectively learning long-term dependencies in sequential data.

[0076] In the disclosed embodiments, after acquiring user behavior data and vehicle environment data, the electronic device can determine a vehicle operating scenario. The vehicle operating scenario is the scenario the vehicle is currently experiencing. The vehicle operating scenario includes multiple scenario attributes, each of which corresponds to one of the user behavior data and vehicle environment data. For example, if the acquired user behavior data includes opening the windows and applying makeup, and the vehicle environment data includes rain, a temperature of 23 degrees Celsius, and traffic congestion, the scenario attributes of the vehicle operating scenario include opening the windows, applying makeup, rain, a temperature of 23 degrees Celsius, and traffic congestion. Together, these scenario attributes constitute the vehicle's current operating scenario. It is understood that user behavior during vehicle operation is highly dynamic and diverse. Even within the same vehicle, on the same road, during the same time period, different users, or even the same user on different days, may exhibit significantly different behaviors. The vehicle's environment is also influenced by multiple factors. Due to the dynamic and diverse nature of user behavior and the complexity of the vehicle environment, the acquired user behavior data and vehicle environment data at each moment in the vehicle's operation exhibit unique characteristics specific to that moment, meaning that each instantaneous scenario is a unique combination. The vehicle operation scenario determined by the embodiment of the present disclosure, which is jointly determined by the user's behavior data and the vehicle environment data during the vehicle operation process, can more carefully and accurately depict the specific operation scenario of the vehicle at the current moment.

[0077] In an embodiment of the present disclosure, after the electronic device determines the vehicle operation scenario based on the behavioral data and the vehicle environment data using the first GRU module, it generates an initial vehicle function service based on the vehicle operation scenario. Among them, the initial vehicle function service generated based on the vehicle operation scenario may include one or more. Exemplarily, after the vehicle recognizes that the current vehicle operation scenario is rainy outside, the temperature is 23 degrees, and the windows are open, the initial vehicle function service is generated based on the vehicle operation scenario. For example, the initial vehicle function service may be adjusting the air-conditioning temperature in the car and / or closing the windows. For another example, when the vehicle recognizes that the current vehicle operation scenario is traffic congestion, rain, and the user in the driver's seat is sleeping, it indicates that there is a safety risk. The initial function service generated based on the vehicle operation scenario may be an automatic driving assistance function service (such as automatic braking, lane keeping assist) and / or music playback service.

[0078] It should be noted that in the vehicle scene recognition method of the related art, the first scene corresponding to the vehicle environment data is usually determined from the mapping relationship based on the vehicle environment data and the preset environment-scene mapping relationship, and the second scene corresponding to the user behavior data is determined from the mapping relationship based on the user behavior data and the preset behavior-scene mapping relationship, and the first scene and the second scene are simply spliced ​​to obtain the current scene of the vehicle, that is, the static mapping method is used to determine the vehicle scene in the related art. The disclosed embodiment introduces a first GRU module as a processing module for scene recognition. With the unique update gate and reset gate mechanism of the GRU module, it can adaptively adjust the information flow based on the correlation between user behavior and vehicle environment through the gating mechanism, which not only retains key data but also effectively filters out irrelevant noise. Compared with the traditional static mapping method, the GRU module can independently mine the deep temporal correlation between behavioral data and environmental data to determine the current vehicle operation scenario, and generate initial vehicle services that conform to the vehicle operation scenario. For example, when the "makeup" behavior and the "rainy day" environment are detected at the same time, the interaction weights of the two types of features can be automatically strengthened, and the current vehicle operation scenario is obtained as "rainy day makeup", and based on this scenario, initial vehicle services that conform to "rainy day makeup" are generated, such as closing windows. The disclosed embodiment adopts the first GRU module to determine the vehicle operation scenario based on behavioral data and vehicle environment data. The method of generating initial vehicle functional services based on the vehicle operation scenario can more carefully and accurately portray the specific operation scenario of the vehicle at the current moment, and generate initial vehicle functional services that conform to the operation scenario based on the current specific operation scenario, so as to improve the accuracy of the determined initial vehicle functional services. In addition, the GRU module has lower computational complexity than other model modules (such as modules in the long short-term memory network model).

[0079] In step S13, the electronic device uses the first interest learning module to adjust the initial vehicle function service based on the preference data to obtain a target vehicle function service. The target vehicle function service may be one or more. For example, when there is one initial vehicle function service, the target vehicle function service obtained by adjusting the initial vehicle function service based on the preference data is also one. When there are multiple initial vehicle function services, the electronic device uses the first interest learning module to adjust each initial vehicle function service based on the preference data to obtain multiple target vehicle function services.

[0080] In the disclosed embodiment, the first GRU module has been utilized through the aforementioned method to determine the vehicle operation scenario based on the behavioral data and the vehicle environment data, and to generate initial vehicle functional services based on the vehicle operation scenario. As previously mentioned, the generated initial vehicle functional services may only include the type of functional service, such as adjusting the air conditioner, opening the window, starting the music player, etc., without the specific configuration parameters of each functional service, such as the target temperature for adjusting the air conditioner, the target degree of opening the window, the type of music to be played after starting the music player, etc. It is understandable that different users generally have different preferences for various functional services. For example, user A likes to adjust the air conditioner to 24 degrees on rainy days, while user B likes to adjust the air conditioner to 26 degrees on rainy days, etc.

[0081] It should be noted that the electronic device uses the first interest learning module to adjust the initial vehicle function service based on the preference data to adjust the configuration parameters of the initial vehicle function service, and can also generate a vehicle function service different from the initial vehicle function service based on the vehicle operation scenario and the user's preference data for the vehicle function service in the vehicle operation scenario, and can also generate another vehicle function service associated with the initial vehicle function service based on the initial vehicle function service and the user's preference data. That is, the first interest learning module in the embodiment of the present disclosure can not only adjust the configuration parameters of the initial vehicle function service, but also generate a vehicle function service different from the initial vehicle function service based on the vehicle operation scenario and the preference data, and use the vehicle function service and the initial vehicle function service together as target vehicle function services.

[0082] For example, in some embodiments, the electronic device uses the first interest learning module to adjust the initial vehicle function service based on the preference data, and the obtained target vehicle function service also includes the type of function service and the specific configuration parameters of the function service. For example, when the initial vehicle function service is to turn on the music player, and the user's preference data for the music playback service is to play rock music, the target vehicle function service is to turn on the music player and automatically play rock music.

[0083] In other embodiments, as mentioned above, the user's preference data for vehicle function services may include the user's preference data for vehicle function services in different vehicle operation scenarios. The electronic device may use the first interest learning module to adjust the initial vehicle function service based on the preference data and generate a vehicle function service different from the initial vehicle function service based on the determined vehicle operation scenario and the user's preference data for vehicle function services in the vehicle operation scenario, and use both the initial vehicle function service and the vehicle function service generated by the first interest learning module that is different from the initial vehicle function service as target vehicle function services. For example, when the user's preference data for vehicle function services includes liking to listen to music while putting on makeup, the vehicle operation scenario is the user putting on makeup on a rainy day, and the initial vehicle function service is closing the windows, the first interest learning module may generate a music playing function service based on the user's preference data for vehicle function services, and simultaneously use closing the windows and playing music as target vehicle function services.

[0084] In other embodiments, the electronic device may use the first interest learning module to adjust the initial vehicle function service based on the preference data to generate another vehicle function service associated with the initial vehicle function service, and use the initial vehicle function service and the other vehicle function service associated with the initial vehicle function service generated by the first interest learning module as target vehicle function services. For example, when the initial vehicle function service is to turn on navigation and the navigation destination is place A, and the user's preference data for the vehicle function service includes that the user is accustomed to buying a cup of coffee at each new place he goes to, the target vehicle function service includes turning on navigation and the navigation destination is place A, and ordering coffee at a coffee shop near place A.

[0085] It can be understood that compared with the method of determining vehicle function services based on the preset mapping relationship between vehicle scenarios and vehicle function services in related technologies, the electronic device in the embodiment of the present disclosure uses the first GRU module to determine the vehicle operation scenario and generate an initial vehicle function service, and then uses the first interest learning module to adjust the initial vehicle function service based on preference data to obtain the target vehicle function service. It can realize the joint generation of predicted vehicle function services by multiple modules, that is, the continuous prediction of the first GRU module and the first interest learning module is realized, which is highly intelligent. Different modules make predictions based on different data, which can also improve the professionalism of each module in making predictions, thereby improving the accuracy of the vehicle function services finally recommended to users.

[0086] In the embodiments of the present disclosure, the electronic device uses the first interest learning module to adjust the initial vehicle function service based on the preference data, obtains the target vehicle function service, and recommends the function service of the vehicle to the user based on the target vehicle function service. In some embodiments, the obtained target function service is one, and the target function service is recommended to the user, such as being recommended in the form of voice broadcast or being displayed on the built-in display screen of the vehicle. In other embodiments, when the target function service is multiple, each target function service can be recommended in a preset order, and each target function service can be sorted according to a preset rule, and the function service recommended to the user is determined from the sorted target function services, such as selecting the first target function service after sorting as the function service recommended to the user.

[0087] In the embodiments of the present disclosure, the electronic device uses the first acquisition module in the trained recommendation model to obtain the vehicle environment data in the vehicle running process, the behavior data of the user of the vehicle, and the preference data of the user for the vehicle function service, and then uses the first GRU module in the trained recommendation model to determine the vehicle running scene based on the behavior data and the vehicle environment data, generate the initial vehicle function service based on the vehicle running scene, and use the first interest learning module in the trained recommendation model to adjust the initial vehicle function service based on the preference data, obtain the target vehicle function service, and recommend the function service of the vehicle to the user based on the target vehicle function service. On the one hand, the recommendation model is a trained model with high accuracy. On the other hand, the present disclosure determines the vehicle running scene based on the user behavior data and the vehicle environment data in the vehicle running process, which can improve the accuracy and uniqueness of the determined vehicle running scene, thereby improving the accuracy of the initial vehicle function service generated based on the vehicle running scene. On the other hand, the present disclosure adjusts the initial vehicle function service based on the preference data of the user for the vehicle function service to obtain the target function service, and then recommends the function service of the vehicle to the user based on the target vehicle function service, which can make the function service recommended to the user more in line with the personalized needs of the user, improve the user experience and satisfaction, and have high intelligence.

[0088] In some embodiments, the training method of the recommendation model comprises:

[0089] The second acquisition module of the preset neural network model is used to obtain a sample data set. The sample data set includes a plurality of time sequence samples, and one time sequence sample includes sample vehicle environment data, sample user behavior data, sample user preference data for vehicle function service, and sample vehicle function service.

[0090] Using the second GRU module of the neural network model, for each time series sample, a correlation relationship is constructed between the sample vehicle environment data and the sample user's behavior data to obtain a sample vehicle operation scenario, and a prediction is performed based on the sample vehicle operation scenario to obtain a sample initial vehicle function service;

[0091] Using the second interest learning module of the neural network model, for each time series sample, based on the sample user's preference data for vehicle function services, adjust the sample initial vehicle function services corresponding to the time series sample to obtain a vehicle function service prediction result for the time series sample;

[0092] Based on the difference between the vehicle function service prediction result of each time series sample and the sample vehicle function service, the parameters of the neural network model are adjusted to obtain the recommendation model.

[0093] In the embodiment of the present disclosure, since the aforementioned recommendation model is a trained model, the preset neural network model needs to be trained before obtaining the trained recommendation model.

[0094] In an embodiment of the present disclosure, an electronic device uses a second acquisition module of a preset neural network model to obtain a sample data set; wherein, the sample data set includes multiple time series samples, and one time series sample includes sample vehicle environment data, sample user behavior data, sample user preference data for vehicle function services, and sample vehicle function services. The comment generation device first obtains a sample training set. In some embodiments, the electronic device uses the second acquisition module of a preset neural network model to obtain historical sample vehicle environment data, sample user historical behavior data, sample user historical preference data for vehicle function services, and historical sample vehicle function services, and performs data processing on each historical data to obtain a sample data set.

[0095] In other embodiments, the electronic device may use a second acquisition module of a preset neural network model to obtain sample vehicle environment data input by the user, sample user behavior data, sample user preference data for vehicle function services, and sample vehicle function services.

[0096] In other embodiments, the electronic device may also obtain sample data sets based on generative models such as generative adversarial networks (GANs) using the second acquisition module of the preset neural network model. For example, GANs may generate sample vehicle environment data, sample user behavior data, and sample user preference data for vehicle function services, and may generate sample vehicle function services based on the sample vehicle environment data, sample user behavior data, and sample user preference data for vehicle function services.

[0097] It should be noted that the data in the sample data set obtained by the electronic device using the second obtaining module of the preset neural network model is already formatted data, and the data in the sample data set is in a format that can be recognized by the model.

[0098] In the embodiments of the present disclosure, the electronic device uses the second GRU module of the neural network model to construct the association relationship between the sample vehicle environment data and the behavior data of the sample user for each time series sample, obtain the sample vehicle running scene, and make a prediction based on the sample vehicle running scene to obtain the sample initial vehicle function service. The electronic device can construct the association relationship between the sample vehicle environment data and the behavior data of the sample user based on the self-attention mechanism. In some embodiments, when the time series sample includes one sample vehicle environment data and one behavior data of the sample user, the self-attention mechanism is used to construct the association relationship between the one sample vehicle environment data and the one behavior data of the sample user.

[0099] In other embodiments, when the time series sample includes multiple sample vehicle environment data and one behavior data of the sample user, the self-attention mechanism is used to construct the association relationship between each sample vehicle environment data and the one behavior data of the sample user, and the sample vehicle running scene is determined based on the multiple association relationships.

[0100] In other embodiments, the construction of the association relationship between the sample vehicle environment data and the behavior data of the sample user and the obtaining of the sample vehicle running scene include:

[0101] When the behavior data of the sample user includes multiple behavior data, the association relationship between each arbitrary two behavior data and the association relationship between each behavior data and the sample vehicle environment data are constructed to obtain the sample vehicle running scene.

[0102] In the embodiments of the present disclosure, the behavior data of the sample user in the time series sample includes multiple behavior data, and the electronic device uses the second GRU module to construct the association relationship between each arbitrary two behavior data and the association relationship between each behavior data and the sample vehicle environment data. For example, the electronic device can use the second GRU module to construct the association relationship between each arbitrary two behavior data and the association relationship between each behavior data and the sample vehicle environment data through the self-attention mechanism. It should be noted that when the sample vehicle environment data also includes multiple sample vehicle environment data, the association relationship between each behavior data and each sample vehicle environment data is constructed, and the sample vehicle running scene is obtained based on the constructed association relationships.

[0103] In the embodiments of the present disclosure, when the user behavior data includes multiple behavior data, the association relationship between behaviors can also be focused on, thereby improving the accuracy of the model prediction.

[0104] In the embodiment of the present disclosure, the electronic device uses the second interest learning module of the neural network model to adjust the sample initial vehicle function service corresponding to each time series sample based on the sample user's preference data for vehicle function service, and obtains the vehicle function service prediction result of the time series sample. The specific process of the electronic device using the second interest learning module of the neural network model to adjust the sample initial vehicle function service corresponding to each time series sample based on the sample user's preference data for vehicle function service, and obtains the vehicle function service prediction result of the time series sample can be similar to the aforementioned process of the electronic device using the first interest learning module to adjust the initial vehicle function service based on the preference data to obtain the target vehicle function service. The embodiment of the present disclosure will not be repeated here.

[0105] In the embodiment of the present disclosure, the electronic device adjusts the parameters of the neural network model based on the difference between the vehicle function service prediction result of each time series sample and the sample vehicle function service to obtain a recommendation model. It should be noted that after the electronic device uses the second acquisition module to obtain the sample data set, it can use the sample data set to iteratively train the second GRU module and the second interest learning module of the neural network model until the convergence condition is met to obtain the recommendation model. Each iterative training process is to use the second GRU module to construct the association relationship between the sample vehicle environment data and the sample user's behavior data for each time series sample, obtain the sample vehicle operation scene, and make a prediction based on the sample vehicle operation scene to obtain the sample initial vehicle function service; use the second interest learning module to adjust the sample initial vehicle function service corresponding to the time series sample based on the sample user's preference data for the vehicle function service for each time series sample, and obtain the vehicle function service prediction result of the time series sample; based on the difference between the vehicle function service prediction result of each time series sample and the sample vehicle function service, fine-tune the parameters of each module in the neural network model to obtain the model parameters after fine-tuning the neural network model.

[0106] In the embodiment of the present disclosure, the convergence condition may be that the difference between the vehicle function service prediction result obtained by the second interest learning module based on the neural network model and the sample vehicle function service is less than a preset difference threshold; the convergence condition may also be that the number of times the neural network model parameters are adjusted is greater than a preset number threshold; the convergence condition may also be that the number of times all sample data in the sample data set are traversed is greater than a preset traversal number threshold, wherein the preset difference threshold, the preset number threshold and the preset traversal number threshold are all set values, such as the preset traversal number threshold can be set to 2, and the embodiment of the present disclosure does not impose any restrictions on this.

[0107] In the disclosed embodiments, it should be noted that the electronic device adjusts the parameters of the neural network model based on the difference between the vehicle function service prediction results of each time series sample and the sample vehicle function service, which may be adjusting the module parameters of at least one module in the neural network model. The second acquisition module after the parameter adjustment is the first acquisition module in the recommendation model, the second GRU module after the parameter adjustment is the first GRU module in the recommendation model, and the second interest learning module after the parameter adjustment is the first interest learning module in the recommendation model.

[0108] In the disclosed embodiment, the parameters of the neural network model are adjusted based on the difference between the vehicle function service prediction results of each time series sample and the sample vehicle function service to obtain a recommendation model, which can reduce the risk of model overfitting, improve the generalization ability of the model, and thereby improve the accuracy of the function services recommended to users by the recommendation model.

[0109] In some embodiments, the step of acquiring the sample data set using the second acquisition module of the preset neural network model includes:

[0110] Utilizing the second acquisition module to acquire a plurality of historical sample vehicle environment data, a plurality of historical sample user behavior data, a plurality of sample user historical preference data for vehicle function services, and a plurality of historical sample vehicle function services;

[0111] Based on preset data processing rules, data segmentation processing is performed on the multiple historical vehicle environment data samples, the multiple historical sample user behavior data, the multiple sample users' historical preference data for vehicle function services, and the multiple historical vehicle function service samples to obtain multiple historical time series samples, and the multiple historical time series samples are used as the sample data set; wherein the data processing rules are used to indicate rules for segmentation into time series samples, and the data processing rules include at least one of the following:

[0112] Classification based on vehicle journey; classification based on the driving time of a vehicle in a trip; classification based on user identity combined with the driving time of a vehicle in a trip.

[0113] As mentioned above, the electronic device can use the second acquisition module of the preset neural network model to obtain historical sample vehicle environment data, historical behavior data of sample users, historical preference data of sample users for vehicle function services, and historical sample vehicle function services, and process each historical data to obtain a sample data set.

[0114] In the embodiment of the present disclosure, the electronic device uses a second acquisition module to acquire a plurality of historical sample vehicle environment data, a plurality of historical sample user behavior data, a plurality of sample user historical preference data for vehicle function services, and a plurality of historical sample vehicle function services. The electronic device may use the second acquisition module to acquire the above historical data by means of a web crawler; the electronic device may also use the second acquisition module to call the above historical data through a predetermined programming interface (Application Programming Interface, API); the electronic device may also use the second acquisition module to acquire the above historical data from a database or data set storing the above historical data, such as a vehicle log data set, through a data processing tool; the electronic device may also establish a communication connection with a vehicle networking platform and use the second acquisition module to acquire the above historical data from the vehicle networking platform. The embodiment of the present disclosure is not limited to this.

[0115] In an embodiment of the present disclosure, the electronic device uses a second acquisition module to perform data division processing on multiple historical vehicle environment data samples, multiple historical sample user behavior data, multiple sample users' historical preference data for vehicle function services, and multiple historical vehicle function service samples based on preset data processing rules to obtain multiple historical time series samples, and use the multiple historical time series samples as a sample data set. In some embodiments, the data processing rules may be based on vehicle trip division. The electronic device may use the second acquisition module to determine the trip corresponding to each data in the above historical data, and determine the data corresponding to the same trip as a time series sample, thereby obtaining multiple time series samples of different trips to form a sample data set. It should be noted that in an embodiment of the present disclosure, the driving process between vehicle ignition and the first shutdown of the vehicle after ignition corresponds to one trip.

[0116] In other embodiments, the data processing rules may be based on the driving time of the vehicle in a trip. The electronic device may use the second acquisition module to determine the trip corresponding to each data in the above historical data, first determine the data corresponding to the same trip, and then determine the timestamp information of the data corresponding to the same trip. Based on the ignition time of the trip, continuous time windows are divided according to a preset fixed time interval (for example, 5 minutes), and all historical data falling into the same time window are determined as a time series sample, thereby obtaining multiple time series samples, each of which represents a data set within a specific time period in a trip.

[0117] In other embodiments, the data processing rules may also be divided based on user identity combined with the driving time of a vehicle in a trip. The electronic device may use the second acquisition module to determine the trip corresponding to each data in the above historical data, first determine the data corresponding to the same trip, then determine the timestamp information of the data corresponding to the same trip and the user identity (such as user ID) corresponding to each data, and divide the continuous time windows according to the preset fixed time interval (for example, 5 minutes) based on the ignition time of the trip, and determine all historical data with the same user identity falling into the same time window as a time series sample, thereby obtaining multiple time series samples, each of which represents a data set of the same user in a specific time period in a trip.

[0118] In the embodiment of the present disclosure, data partitioning and processing are performed on multiple historical vehicle environment data samples, multiple historical sample user behavior data, multiple sample users' historical preference data for vehicle function services, and multiple historical vehicle function service samples based on preset data processing rules to obtain multiple historical time series samples, thereby forming a sample data set. This method can assist in subsequent data analysis of the model, allowing the model to more easily capture the correlation between the data, thereby improving the accuracy of the model prediction.

[0119] In some embodiments, the target vehicle functional services obtained by the first interest learning module include multiple ones, the recommendation model further includes an intention reasoning module, and the method further includes:

[0120] The intention reasoning module is used to sort the multiple target functional services based on preset intervention rules, and a functional service recommended to the user is determined from the sorted multiple target functional services.

[0121] In the embodiment of the present disclosure, the recommendation model also includes an intention reasoning module. It should be noted that the neural network model used for training also includes a second intention reasoning module. The intention reasoning module in the recommendation model is obtained by training the neural network model and adjusting the parameters.

[0122] In the embodiment of the present disclosure, the target vehicle function services obtained by the first interest learning module include multiple, and the electronic device uses the intention reasoning module to sort the multiple target function services based on the preset intervention rules, and determines the function services recommended to the user from the sorted multiple target function services, wherein the preset intervention rules can be pre-entered by the user or set by the developer, and the embodiment of the present disclosure does not limit this. The preset intervention rules may include at least one of the following: giving priority to recommending functions related to safe driving; when the target function service includes recommending to go to different gas stations for refueling, giving priority to recommending to go to gas station A (intervention with operational nature), giving priority to navigation for commercial shopping on holidays (intervention with time nature), giving priority to navigation to check-in outlets for tourism (intervention with LBS nature), etc.

[0123] In an embodiment of the present disclosure, the electronic device uses an intention reasoning module to sort multiple target function services based on a preset intervention rule. The function service recommended to the user from the sorted multiple target function services may include one or more, and the embodiment of the present disclosure does not limit this. For example, the target function service with the highest priority after sorting can be used as the function service recommended to the user, or a preset number of target function services with high priority can be selected from the sorted multiple target function services according to a preset number (for example, three), and the preset number of target function services are recommended to the user in order of priority.

[0124] In the embodiment of the present disclosure, multiple target functional services are ranked and recommended according to preset intervention rules, which can further improve the accuracy and effectiveness of the recommendations.

[0125] In some embodiments, the method further comprises:

[0126] Obtaining feedback from the user on the recommended functional service;

[0127] Parameters of the recommendation model are adjusted based on the vehicle environment data, the behavior data of the user of the vehicle, the preference data of the user for vehicle function services, and the feedback.

[0128] In an embodiment of the present disclosure, the electronic device can also obtain user feedback on the recommended functional services. In some embodiments, if the method of recommending functional services to the user is voice recommendation, for example, the functional service recommended to the user is to adjust the air conditioner to 23 degrees, and the recommendation method is to voice broadcast "It is recommended to adjust the air conditioner to 23 degrees". It can be monitored whether the user accepts the suggestion and adjusts the air conditioner to 23 degrees within a preset time (such as 30 seconds) after the voice broadcast to determine the user's feedback on the recommended functional service; for another example, the functional service recommended to the user is to adjust the air conditioner to 23 degrees, and the recommendation method is to voice broadcast "Do you need to adjust the air conditioner to 23 degrees?" It can be monitored whether the user responds to the suggestion within a preset time (such as 30 seconds) after the voice broadcast, such as whether the user responds to the suggestion, such as whether the user responds "yes", "no", "okay", "okay", "no", etc. to indicate the user's feedback on the suggestion.

[0129] In other embodiments, if the method of recommending a functional service to the user is to display a recommendation, for example, the functional service recommended to the user is to adjust the air conditioner to 23 degrees. The recommendation method is to display "It is recommended to adjust the air conditioner to 23 degrees" on the vehicle display screen. It can be monitored within a preset time (such as 30 seconds) after the functional service is displayed whether the user accepts the suggestion and adjusts the air conditioner to 23 degrees to determine the user's feedback on the recommended functional service; for another example, the functional service recommended to the user is to adjust the air conditioner to 23 degrees. The recommendation method is to display "Do you need to adjust the air conditioner to 23 degrees" on the vehicle display screen. It can be monitored within a preset time (such as 30 seconds) after the display to determine the user's response to the suggestion, such as whether the user responds with "yes", "no", "okay", "okay", "no" or the like to indicate the user's feedback on the suggestion, or two decision windows can be displayed when the functional service is displayed, one decision window displays "yes" and the other decision window displays "no". The user's selection operation on the decision window within a preset time (such as 30 seconds) after the display can be monitored to determine the user's feedback on the recommended functional service.

[0130] In an embodiment of the present disclosure, the electronic device adjusts the parameters of the recommendation model based on vehicle environmental data, vehicle user behavior data, user preference data for vehicle functional services, and feedback. For example, when the user's feedback on the recommended functional service is to accept the recommendation, the parameters of the current recommendation model can be maintained. When the user's feedback on the recommended functional service is to reject the recommendation, the parameters of the recommendation model can be adjusted based on the vehicle environmental data, vehicle user behavior data, user preference data for vehicle functional services, and feedback. More detailed adjustments can also be made based on the user's behavior after rejecting the recommendation. For example, if the functional service recommended to the user is to adjust the air conditioner to 23 degrees, the user rejects the recommendation but adjusts the air conditioner to 26 degrees. This indicates that the type of the recommended functional service is correct but the specific parameters are incorrect. At this time, only the parameters of the first interest learning module and / or intention reasoning module in the recommendation model can be adjusted.

[0131] The disclosed embodiment can further adjust the model parameters based on user feedback and the recommended vehicle environment data, user behavior data, and user preference data for vehicle function services, so that the model is dynamically adjusted during use, with high intelligence, and real-time adjustment of the model can also continuously improve the accuracy of model recommendations.

[0132] Figure 2 is a flowchart illustrating a recommendation method according to an exemplary embodiment. Figure 2 As shown, the method includes the following steps:

[0133] S21. Obtain target data and convert the target data into a vector representation. The target data includes first data, second data, and third data. The first data is the behavioral data of existing vehicle users required by the conversation data model, the second data is the portrait data and environmental features required for user portraits, and the third data is the interaction feedback data.

[0134] In the embodiment of the present disclosure, the first data is the user's behavioral data, the second data is the user's preference data for vehicle functional services and external environment data in vehicle environmental data, and the third data is the data generated by the user's interaction with various vehicle functional services. It should be noted that this process is a training process, and the above-mentioned data obtained are all historical data. The target data is obtained using the second acquisition module of the preset neural network model.

[0135] S22: splicing the target data to obtain fused data.

[0136] In the disclosed embodiment, the aforementioned first data, second data, and third data are simply concatenated to obtain fused data.

[0137] S23. Acquire a conversation model based on the fused data.

[0138] In the embodiment of the present disclosure, taking the conversation model located in the second acquisition module as an example, step S23 refers to performing data division processing on the above-mentioned fusion data based on preset data processing rules to obtain multiple historical time series samples, and using the multiple historical time series samples as sample data sets.

[0139] S24. Input the data in the session data model into a fusion decision model for training. The architecture of the fusion decision model includes a GRU module, an interest learning module, and an intention reasoning module.

[0140] In the embodiment of the present disclosure, the fusion decision model here is a preset neural network model, the GRU module is the second GRU module, the interest learning module is the second interest learning module, and the intention reasoning module is the intention reasoning module in the untrained neural network model. This process is the process of training the preset neural network model based on the sample data set.

[0141] S25: Train and evaluate the fusion decision model, and adjust model parameters based on training feedback.

[0142] In the embodiment of the present disclosure, the difference between the vehicle function service prediction result based on each historical time series sample and the sample vehicle function service (wherein the first data includes the sample vehicle function service) is characterized to adjust the parameters of the neural network model.

[0143] S26. Based on the trained fusion decision model, obtain the real-time data of the user's vehicle and generate scene recommendations in a timely manner.

[0144] In the embodiment of the present disclosure, taking the trained fusion decision model as the aforementioned trained recommendation model as an example, the trained fusion decision model is used to obtain real-time data of the user's vehicle and generate scene recommendations at the right time, that is, to obtain vehicle environment data during vehicle operation, vehicle user behavior data, and user preference data for vehicle functional services, and recommend vehicle functional services to users based on the obtained data.

[0145] Figure 3 is an architectural diagram of a recommendation model according to an exemplary embodiment. Figure 3As shown, L31 is an embedding layer, which is used to convert the basic behaviors of users in the vehicle (such as the user's operation behaviors on vehicle services, multimedia, voice, mini-programs, maps, vehicle control and vehicle settings, etc.) into low-dimensional vector representations. L32 is an embedding layer, which is used to convert the user's basic portrait data L321 and the environmental feature data L322 during vehicle operation into vector representations. L33 is an embedding layer, which is used to convert the interactive feedback data between the user in the vehicle and the vehicle function service into a vector representation. L34 is a splicing layer, which is used to splice the aforementioned vector representations to form a more comprehensive feature vector. L35 is an automatic feature learning converter, which is used to align and convert the acquired feature data (such as user behavior data, vehicle environment data), and learn the complex time series and dependencies contained in these sequence data through the "converter layer". L36 is the object to be recommended, that is, the process of determining the vehicle operation scenario based on the acquired user behavior data and vehicle environment data, and determining the initial vehicle function service based on the vehicle operation scenario. The object to be recommended is the initial vehicle function service. The gated recurrent unit L38 is the first GRU module. As can be seen from the figure, the gated recurrent unit L38 constructs the association relationship in the acquired data, determines the vehicle operation scenario, and generates the initial vehicle function service. L37 is used to receive the feature vectors from the embedding layer L32 and the embedding layer L33, and perform embedding fusion, and use the deep neural network model to perform "self-attention" mechanism and "path allocation" to deeply mine and learn the user's preference data for vehicle function services, and train through "contrast loss". That is, L37 can learn the user's potential and deep preference for vehicle function services from complex historical behavior sequences and user basic features (such as user basic portraits). Interest learning L39 is the first interest learning module. As can be seen from the figure, interest learning L39 can perform interest mining and interest prediction, and can adjust the initial vehicle function service generated by the gated recurrent unit L38 based on the user's preference data for vehicle function services determined in L37 to obtain the target vehicle function service. Intent reasoning L40 is the intention reasoning module in the recommendation model. As can be seen from the figure, intention reasoning L40 can perform scene intention intervention and intention sorting on the target vehicle functional services obtained by interest learning L39, so as to determine the functional services recommended to users from the sorted multiple target functional services and make recommendations.

[0146] Figure 4 4 is a block diagram of a recommendation device 400 according to an exemplary embodiment. The recommendation model after training is applied, and the recommendation model includes a first acquisition module, a first gated recurrent unit GRU module, and a first interest learning module, such as Figure 4 As shown, the device mainly includes:

[0147] The third obtaining module 401 is configured to obtain, by using the first obtaining module, vehicle environment data in a vehicle operation process, behavior data of a user of the vehicle, and preference data of the user for a vehicle function service;

[0148] The generating module 402 is configured to determine, by using the first GRU module, a vehicle operation scenario based on the behavior data and the vehicle environment data, and generate an initial vehicle function service based on the vehicle operation scenario.

[0149] The recommending module 403 is configured to adjust, by using the first interest learning module, the initial vehicle function service based on the preference data to obtain a target vehicle function service to be recommended, and recommend, based on the target vehicle function service, the function service of the vehicle for the user.

[0150] In some embodiments, the apparatus further includes:

[0151] The recommendation model training module is configured to obtain, by using a second obtaining module of a preset neural network model, a sample data set, wherein the sample data set includes a plurality of time series samples, and each time series sample includes sample vehicle environment data, sample user behavior data, sample user preference data for a vehicle function service, and a sample vehicle function service; for each time series sample, a second GRU module of the neural network model is used to construct an association relationship between the sample vehicle environment data and the sample user behavior data, to obtain a sample vehicle operation scenario, and based on the sample vehicle operation scenario, a prediction is made to obtain a sample initial vehicle function service; for each time series sample, a second interest learning module of the neural network model is used to adjust the sample initial vehicle function service corresponding to the time series sample based on the sample user preference data for the vehicle function service, to obtain a vehicle function service prediction result of the time series sample; based on the difference between the vehicle function service prediction result of each time series sample and the sample vehicle function service, the parameters of the neural network model are adjusted to obtain the recommendation model.

[0152] In some embodiments, the recommendation model training module is further configured to, in a case where the sample user behavior data includes a plurality of behavior data, construct an association relationship between each of the behavior data and an association relationship between each behavior data and the sample vehicle environment data, to obtain the sample vehicle operation scenario.

[0153] In some embodiments, the recommendation model training module is further configured to use the second acquisition module to obtain multiple historical sample vehicle environment data, multiple historical sample user behavior data, multiple sample users' historical preference data for vehicle function services, and multiple historical sample vehicle function services; based on preset data processing rules, the multiple historical vehicle environment data samples, the multiple historical sample user behavior data, the multiple sample users' historical preference data for vehicle function services, and the multiple historical vehicle function service samples are subjected to data division processing to obtain multiple historical time series samples, and the multiple historical time series samples are used as the sample data set; wherein, the data processing rules are used to indicate the rules for division into time series samples, and the data processing rules include at least one of the following: division based on vehicle trip; division based on the driving time of the vehicle in a trip; division based on user identity combined with the driving time of the vehicle in a trip.

[0154] In some embodiments, the apparatus further comprises:

[0155] The preference data determination module is configured to determine the user's preference data for vehicle function services in a method including at least one of the following: determining based on the user's identity information data; determining based on the user's historical interaction data with the vehicle function services.

[0156] In some embodiments, the target vehicle functional services obtained by the first interest learning module include multiple ones, the recommendation model further includes an intention reasoning module, and the device further includes:

[0157] The function service determination module is configured to use the intention reasoning module to sort the multiple target function services based on preset intervention rules, and determine the function service recommended to the user from the sorted multiple target function services.

[0158] In some embodiments, the apparatus further comprises:

[0159] a fourth acquisition module, configured to obtain feedback from the user on the recommended functional service;

[0160] An adjustment module is configured to adjust the parameters of the recommendation model based on the vehicle environment data, the behavior data of the user of the vehicle, the preference data of the user for vehicle function services and the feedback.

[0161] Regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0162] Figure 5This is a block diagram illustrating the structure of an electronic device 500 according to an exemplary embodiment. For example, electronic device 500 may be a terminal device such as a user device, mobile device, tablet computer, personal digital assistant, handheld device, computing device, vehicle-mounted device, wearable device, etc., which has a built-in trained recommendation model. It may also be a cloud device such as a cloud server, which has a built-in trained recommendation model. This is not limited in the present embodiment.

[0163] Reference Figure 5 , electronic device 500 may include one or more of the following components: a processing component 502 , a memory 504 , a power component 506 , a multimedia component 508 , an audio component 510 , an input / output (I / O) interface 512 , a sensor component 514 , and a communication component 516 .

[0164] The processing component 502 generally controls the overall operation of the electronic device 500, such as operations associated with at least one of display, phone calls, data communications, camera operation, and recording operations. The processing component 502 may include one or more processors 520 to execute instructions to perform all or part of the steps of the above-described method. In addition, the processing component 502 may include one or more modules to facilitate interaction between the processing component 502 and other components. For example, the processing component 502 may include a multimedia module to facilitate interaction between the multimedia component 508 and the processing component 502.

[0165] The memory 504 is configured to store various types of data to support operations on the electronic device 500. Examples of such data include at least one of the following: instructions for any application or method operating on the electronic device 500, contact data, phone book data, messages, pictures, and videos. The memory 504 can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0166] The power supply component 506 provides power to various components of the electronic device 500. The power supply component 506 may include at least one of the following: a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device 500.

[0167] The multimedia component 508 includes a screen that provides an output interface between the electronic device 500 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, slides, and gestures on the touch panel. The touch sensor can not only sense the boundaries of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 508 includes a front camera and / or a rear camera. When the electronic device 500 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have focal length and optical zoom capabilities.

[0168] The audio component 510 is configured to output and / or input audio signals. For example, the audio component 510 includes a microphone (MIC), which is configured to receive external audio signals when the electronic device 500 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signal can be further stored in the memory 504 or transmitted via the communication component 516. In some embodiments, the audio component 510 also includes a speaker for outputting audio signals.

[0169] I / O interface 512 provides an interface between processing component 502 and peripheral interface modules, such as a keyboard, click wheel, and buttons. These buttons may include, but are not limited to, a home button, volume buttons, a start button, and a lock button.

[0170] The sensor component 514 includes one or more sensors for providing status assessments for various aspects of the electronic device 500. For example, the sensor component 514 can detect an open / closed position of the electronic device 500, relative positioning of components, such as a display and a keypad of the electronic device 500, a change in position of the electronic device 500 or a component of the electronic device 500, presence or absence of user contact with the electronic device 500, orientation or acceleration / deceleration / rotation of the electronic device 500, and temperature changes of the electronic device 500. The sensor component 514 can include an accelerometer to detect a change in position, orientation, and other inertial measurements. The sensor component 514 can also include a proximity sensor configured to detect presence of a nearby object without any physical touch. The sensor component 514 can further include a light sensor, such as a complementary metal-oxide-semiconductor (CMOS) or charge-coupled device (CCD) image sensor, utilized in an imaging application. In some embodiments, the sensor component 514 can further include at least one of an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, and a temperature sensor, among others.

[0171] The communication component 516 is configured to facilitate wired or wireless communication between the electronic device 500 and other devices. The electronic device 500 can access a wireless network based on a communication standard, such as Wi-Fi, 4G, 5G, or a combination thereof. In an example embodiment, the communication component 516 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In an example embodiment, the communication component 516 can further include a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra Wide Band (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0172] In an exemplary embodiment, the electronic device 500 may be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components.

[0173] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is further provided, such as a memory 504 including executable instructions or a computer program. The instructions or computer program can be executed by the processor 520 of the electronic device 500 to perform the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, etc.

[0174] A non-transitory computer-readable storage medium, when the instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to perform any of the above-mentioned recommended methods in the embodiments of the present disclosure.

[0175] The present disclosure provides a computer program product comprising a computer program or executable instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer program or executable instructions from the computer-readable storage medium and executes the computer program or executable instructions, causing the computer device to perform any of the aforementioned recommendation methods of the present disclosure.

[0176] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the claims.

[0177] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A recommendation method, characterized in that: The method is applied to a trained recommendation model, wherein the recommendation model includes a first acquisition module, a first gated recurrent unit (GRU) module, and a first interest learning module. The method includes: Using the first acquisition module to acquire vehicle environment data during vehicle operation, behavior data of a user of the vehicle, and preference data of the user for vehicle function services; Determine, using the first GRU module, a vehicle operation scenario based on the behavior data and the vehicle environment data, and generate an initial vehicle function service based on the vehicle operation scenario; The first interest learning module is used to adjust the initial vehicle function service based on the preference data to obtain a target vehicle function service, and the vehicle function service is recommended to the user based on the target vehicle function service.

2. The recommendation method according to claim 1, characterized in that The training method of the recommendation model includes: A sample data set is acquired using a second acquisition module of a preset neural network model; wherein the sample data set includes a plurality of time series samples, and each time series sample includes sample vehicle environment data, sample user behavior data, sample user preference data for vehicle function services, and sample vehicle function services; Using the second GRU module of the neural network model, for each time series sample, a correlation relationship is constructed between the sample vehicle environment data and the sample user's behavior data to obtain a sample vehicle operation scenario, and a prediction is performed based on the sample vehicle operation scenario to obtain a sample initial vehicle function service; Using the second interest learning module of the neural network model, for each time series sample, based on the sample user's preference data for vehicle function services, adjust the sample initial vehicle function services corresponding to the time series sample to obtain a vehicle function service prediction result for the time series sample; Based on the difference between the vehicle function service prediction result of each time series sample and the sample vehicle function service, the parameters of the neural network model are adjusted to obtain the recommendation model.

3. The method according to claim 2, characterized in that The step of constructing a correlation between the sample vehicle environment data and the sample user behavior data to obtain a sample vehicle operation scenario includes: In the case where the sample user's behavior data includes multiple items, an association relationship between any two pieces of behavior data and an association relationship between each piece of behavior data and the sample vehicle environment data are constructed to obtain the sample vehicle operation scenario.

4. The method according to claim 2 or 3, characterized in that The method of acquiring the sample data set by using the second acquisition module of the preset neural network model includes: Utilizing the second acquisition module to acquire a plurality of historical sample vehicle environment data, a plurality of historical sample user behavior data, a plurality of sample user historical preference data for vehicle function services, and a plurality of historical sample vehicle function services; Based on preset data processing rules, data segmentation processing is performed on the multiple historical vehicle environment data samples, the multiple historical sample user behavior data, the multiple sample users' historical preference data for vehicle function services, and the multiple historical vehicle function service samples to obtain multiple historical time series samples, and the multiple historical time series samples are used as the sample data set; wherein the data processing rules are used to indicate rules for segmentation into time series samples, and the data processing rules include at least one of the following: Classification based on vehicle journey; classification based on the driving time of a vehicle in a trip; classification based on user identity combined with the driving time of a vehicle in a trip.

5. The method according to claim 1, wherein The method for determining the user's preference data for vehicle function services includes at least one of the following: Determined based on the user's identity information data; Determined based on the user's historical interaction data with vehicle function services.

6. The method according to claim 1, characterized in that The target vehicle functional services obtained by the first interest learning module include multiple ones, the recommendation model further includes an intention reasoning module, and the method further includes: The intention reasoning module is used to sort the multiple target functional services based on preset intervention rules, and a functional service recommended to the user is determined from the sorted multiple target functional services.

7. The method according to claim 1, characterized in that The method further comprises: Obtaining feedback from the user on the recommended functional service; Parameters of the recommendation model are adjusted based on the vehicle environment data, the behavior data of the user of the vehicle, the preference data of the user for vehicle function services, and the feedback.

8. A recommendation device, characterized in that: The recommendation model is applied to the trained recommendation model, the recommendation model includes a first acquisition module, a first gated recurrent unit (GRU) module, and a first interest learning module, and the device includes: a third acquisition module configured to use the first acquisition module to acquire vehicle environment data during vehicle operation, behavior data of a user of the vehicle, and preference data of the user for vehicle function services; a generation module configured to, using the first GRU module, determine a vehicle operation scenario based on the behavior data and the vehicle environment data, and generate an initial vehicle function service based on the vehicle operation scenario; The recommendation module is configured to use the first interest learning module to adjust the initial vehicle function service based on the preference data to obtain the target vehicle function service to be recommended, and recommend vehicle function services to the user based on the target vehicle function service.

9. An electronic device, characterized in that: include: processor; memory for storing computer programs or instructions; The processor executes the computer program or instructions to implement the steps of the recommendation method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing a computer program or instruction, characterized in that: When the computer program or instructions in the storage medium are executed by a processor, the steps of the recommendation method according to any one of claims 1 to 7 are implemented.

11. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the recommendation method according to any one of claims 1 to 7 are implemented.

Images