Parking space recommendation method, device, equipment, storage medium and product

Abstract

The application discloses a parking space recommendation method and device, equipment, storage medium and product, and relates to the technical field of new energy vehicles. The parking space recommendation method comprises the following steps: receiving a parking instruction, determining a parking demand based on the parking instruction, and performing multi-dimensional data matching based on the parking demand to obtain a parking space recommendation result. The multi-dimensional data comprises multiple items in static map data, user preference data and real-time dynamic data. Since the multi-dimensional data of the static map data, the user preference data and the real-time dynamic data is comprehensively considered in the parking space recommendation, the parking space recommendation result can be more in line with the user demand, and the parking efficiency and the user experience are improved.

Description

Technical Field

[0001] This application relates to the field of new energy vehicle technology, and in particular to a parking space recommendation method, device, equipment, storage medium and product. Background Technology

[0002] In related technologies, with the continuous increase in car ownership, the problem of parking difficulties has become increasingly prominent. Existing intelligent parking systems have a simplistic logic when recommending parking spaces, only considering basic information such as whether the parking space is vacant. This often results in recommended parking spaces not meeting user needs, reducing parking efficiency and user experience. Summary of the Invention

[0003] The main purpose of this application is to provide a parking space recommendation method, device, equipment, storage medium and product, which aims to solve the technical problem that considering only basic information such as whether the parking space is vacant reduces parking efficiency and user experience.

[0004] To achieve the above objectives, this application proposes a parking space recommendation method, which includes:

[0005] Receive parking instructions and determine parking requirements based on the parking instructions;

[0006] Based on the parking demand, multi-dimensional data matching is performed to obtain parking space recommendation results. The multi-dimensional data includes multiple items from static map data, user preference data, and real-time dynamic data.

[0007] In one embodiment, the step of receiving a parking instruction and determining a parking requirement based on the parking instruction includes, prior to:

[0008] When the vehicle is in a preset parking environment, based on the scene accuracy of the preset parking environment, static map data and user preference data corresponding to the scene accuracy are loaded, and real-time dynamic data is collected. The static map data includes at least one of real-time modeled map and structured map, and the user preference data includes at least one of general historical preference data and specific garage historical data.

[0009] In one embodiment, the step of performing multi-dimensional data matching based on the parking demand to obtain parking space recommendation results includes:

[0010] Determine the recommendation dimension corresponding to the parking demand, wherein the recommendation dimension includes any one of the basic recommendation dimension and the combined recommendation dimension;

[0011] From the parking space nodes in the preset semantic graph, the target parking space nodes corresponding to the recommendation dimension are selected, and the parking space recommendation results are determined based on the target parking space nodes.

[0012] In one embodiment, the step of performing multi-dimensional data matching based on the parking demand to obtain parking space recommendation results includes:

[0013] Acquire multi-dimensional data and convert the multi-dimensional data into semantic graph nodes, wherein the semantic graph nodes include scene nodes, reference target nodes, parking space attribute nodes, user preference nodes, and parking space nodes;

[0014] The association weights between nodes in the semantic graph are quantified, and node connections between the semantic graph nodes are established based on the association weights to obtain a preset semantic graph.

[0015] In one embodiment, the association weights include scene association weights, and the step of quantifying the association weights between nodes in the semantic graph includes:

[0016] Calculate the straight-line distance from each parking space corresponding to the parking space node to the key facility corresponding to the scene node;

[0017] The number of times each parking space is entered in the scene corresponding to each scene node is counted;

[0018] Based on the straight-line distance and the number of parking entries, the scene association weight between the parking space node and the scene node in the semantic graph is quantified.

[0019] In one embodiment, the association weights include target association weights, and the step of quantifying the association weights between nodes in the semantic graph further includes:

[0020] Calculate the straight-line distance from each parking space corresponding to the parking space node to each reference target of the reference target node;

[0021] Based on the straight-line distance, the target association weights between the parking space node and the scene node in the semantic graph are quantified for different semantic types, wherein the semantic types include tendency type and avoidance type.

[0022] In one embodiment, the association weights include attribute association weights, and the step of quantifying the association weights between nodes in the semantic graph further includes:

[0023] Based on the multi-dimensional data, the matching degree between each parking space corresponding to the parking space node and the parking space attribute corresponding to the parking space attribute node is quantified.

[0024] Based on the matching degree, the attribute association weight between the parking space node and the parking space attribute node in the semantic graph is quantified.

[0025] In one embodiment, the association weights include preference association weights, and the step of quantifying the association weights between nodes in the semantic graph further includes:

[0026] Based on the number of positive and negative feedbacks from the user for each parking space corresponding to the parking space node, a specific preference matching score is calculated.

[0027] Based on the preference weights of the general preference data in the multi-dimensional data, the general preference matching score of each parking space that satisfies the general preference is calculated;

[0028] Based on the specific preference matching score and the general preference matching score, the preference association weight between the parking space node and the user preference node in the semantic graph is quantified.

[0029] In one embodiment, the step of filtering target parking space nodes corresponding to the recommendation dimension from parking space nodes in a preset semantic graph, and determining the parking space recommendation result based on the target parking space nodes, includes:

[0030] Based on the recommendation dimension, the parking space nodes in the preset semantic graph are filtered and the association weight is corrected to obtain candidate parking space nodes.

[0031] Target parking space nodes with a parking difficulty coefficient less than a first threshold are selected from the candidate parking space nodes;

[0032] The target parking space nodes are sorted and output based on the corrected association weights to obtain the parking space recommendation results.

[0033] In one embodiment, the basic recommendation dimension includes scene-based automatic matching recommendation, and the step of filtering and correcting the association weights of parking space nodes in the preset semantic graph based on the recommendation dimension to obtain candidate parking space nodes includes:

[0034] The scene nodes corresponding to the scene are automatically matched and recommended from the nodes of the preset semantic graph;

[0035] Traverse the first association edge between the scene node and the parking space node, and filter out candidate parking space nodes whose scene association weight of the first association edge is greater than or equal to the second threshold.

[0036] Based on the preference association weights corresponding to the scene preference nodes in the user preference nodes, the scene association weights of the candidate parking space nodes are corrected to obtain the corrected candidate parking space nodes.

[0037] In one embodiment, the basic recommendation dimension includes reference target-oriented recommendation, and the step of filtering and correcting the association weights of parking space nodes in the preset semantic graph based on the recommendation dimension to obtain candidate parking space nodes further includes:

[0038] The reference target node corresponding to the reference target-oriented recommendation is determined from the nodes of the preset semantic graph;

[0039] Traverse the second association edge between the reference target node and the parking space node to filter out candidate parking space nodes that meet the distance requirements;

[0040] Based on the preference association weight of the location avoidance preference node in the user preference node, the target association weight of the candidate parking space node is corrected to obtain the corrected candidate parking space node.

[0041] In one embodiment, the basic recommendation dimension includes parking space attribute adaptation recommendation, and the step of filtering and correcting the association weights of parking space nodes in the preset semantic graph based on the recommendation dimension to obtain candidate parking space nodes further includes:

[0042] The parking space attribute nodes corresponding to the parking space attribute adaptation recommendation are determined from the nodes of the preset semantic graph;

[0043] Traverse the third association edge between the parking space attribute node and the parking space node, and filter out candidate parking space nodes whose attribute association weight of the third association edge is greater than the third threshold.

[0044] Based on the preference association weights of the attribute demand preference nodes in the user preference nodes, the attribute association weights of the candidate parking space nodes are corrected to obtain the corrected candidate parking space nodes.

[0045] In one embodiment, the step of filtering parking space nodes in the preset semantic graph based on the recommendation dimension and correcting the association weights to obtain candidate parking space nodes further includes:

[0046] From the nodes of the preset semantic graph, determine the single-dimensional nodes corresponding to the single recommendation dimension in the combined recommendation dimensions;

[0047] The fourth association edge between the single-dimensional node and the parking space node is traversed respectively. Based on the semantic type in the parking requirement, candidate parking space nodes that meet the parking requirement are selected.

[0048] Based on the association weights of each node in the user preference node, the association weights of candidate parking space nodes in different dimensions are corrected, and the corrected association weights are summed in multiple dimensions to obtain the candidate parking space nodes after weighted summation.

[0049] In one embodiment, the step of receiving a parking instruction includes, prior to:

[0050] When a user's voice information is received, the voice information is parsed into semantic demand information to obtain demand information, and a parking instruction is generated based on the demand information. The demand information includes at least scene information, attribute information and target information.

[0051] or,

[0052] When the user does not issue a parking instruction, the system automatically determines the demand information, including the scene information, based on the navigation information, and generates the parking instruction based on the demand information.

[0053] In one embodiment, the step of receiving a parking instruction and determining a parking requirement based on the parking instruction includes:

[0054] The number of times the data attributes of the vehicle parking process overlap with the data attributes of the general historical preference data is calculated, wherein the behavioral data includes at least one of active selection data, active feedback data, and system inference data;

[0055] Based on the data type of the general historical preference data, the number of overlaps is classified to obtain the target number of overlaps corresponding to different data types;

[0056] Based on the number of times the target overlaps and the differentiated weights corresponding to different data types, the preference scores of the general preference data for different data types are calculated.

[0057] Based on the preference scores, the updated dynamic weights of the general preference data for different data types are calculated, wherein the dynamic weights are used to update the preference weights of the general preference data.

[0058] Furthermore, to achieve the above objectives, this application also proposes a parking space recommendation device, which includes:

[0059] The response module is used to receive parking instructions and determine parking needs based on the parking instructions;

[0060] The matching module is used to perform multi-dimensional data matching based on the parking demand to obtain parking space recommendation results. The multi-dimensional data includes multiple items from static map data, user preference data, and real-time dynamic data.

[0061] In addition, to achieve the above objectives, this application also proposes a parking space recommendation device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the parking space recommendation method as described above.

[0062] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and which, when executed by a processor, implements the steps of the parking space recommendation method described above.

[0063] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the parking space recommendation method described above.

[0064] One or more technical solutions proposed in this application have at least the following technical effects:

[0065] In contrast to related technologies, with the continuous increase in car ownership, the parking problem is becoming increasingly prominent. Existing intelligent parking systems rely on a single logic in recommending parking spaces, considering only basic information such as whether a parking space is available. This often results in recommended parking spaces that do not meet user needs, reducing parking efficiency and user experience. This application, however, receives a parking instruction, determines parking requirements based on the instruction, and performs multi-dimensional data matching based on the parking requirements to obtain parking space recommendation results. The multi-dimensional data includes static map data, user preference data, and real-time dynamic data. This application, after determining parking requirements based on a user-triggered parking instruction and recommending parking spaces accordingly, comprehensively considers multi-dimensional data including static map data, user preference data, and real-time dynamic data. This makes the recommended parking spaces more aligned with user needs, thereby improving parking efficiency and user experience. Attached Figure Description

[0066] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0067] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0068] Figure 1 This is a flowchart illustrating the parking space recommendation method of this application in Implementation Example 1.

[0069] Figure 2 This is a flowchart illustrating the intelligent parking space recommendation method of the parking space recommendation method described in this application;

[0070] Figure 3 This is a flowchart illustrating Embodiment 2 of the parking space recommendation method in this application;

[0071] Figure 4 This is a flowchart illustrating Embodiment 3 of the parking space recommendation method of this application;

[0072] Figure 5 This is a flowchart illustrating Embodiment 5 of the parking space recommendation method in this application;

[0073] Figure 6 This is a diagram illustrating the display behavior data acquisition prompts for the parking space recommendation method in this application.

[0074] Figure 7 This is a schematic diagram of the module structure of the parking space recommendation device according to an embodiment of this application;

[0075] Figure 8 This is a schematic diagram of the device structure of the hardware operating environment involved in the parking space recommendation method in this application embodiment.

[0076] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0077] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0078] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0079] The main solution of this application embodiment is: receiving a parking instruction, determining parking demand based on the parking instruction, and performing multi-dimensional data matching based on the parking demand to obtain parking space recommendation results, wherein the multi-dimensional data includes multiple items from static map data, user preference data, and real-time dynamic data.

[0080] In related technologies, with the continuous increase in car ownership, the problem of parking difficulties has become increasingly prominent. Existing intelligent parking systems have a simplistic logic when recommending parking spaces, only considering basic information such as whether the parking space is vacant. This often results in recommended parking spaces not meeting user needs, reducing parking efficiency and user experience.

[0081] This application, after determining parking needs based on parking commands triggered by the user, comprehensively considers multi-dimensional data, including static map data, user preference data, and real-time dynamic data, when recommending parking spaces based on these needs. This makes the parking space recommendation results more in line with user needs, thereby improving parking efficiency and user experience.

[0082] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or parking space recommendation device capable of performing the above functions. The following description uses a parking space recommendation device as an example to illustrate this embodiment and the subsequent embodiments.

[0083] Based on this, this application provides a parking space recommendation method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the parking space recommendation method of this application.

[0084] In this embodiment, the parking space recommendation method includes steps S10 to S20:

[0085] Step S10: Receive parking instruction and determine parking requirements based on the parking instruction;

[0086] It should be noted that the executing entity in this embodiment is the parking space recommendation device. A parking command refers to a command triggered by the user through voice input or touch operation of the vehicle's human-machine interface (HMI), or it can be automatically triggered based on navigation data even if the user does not issue a command. Parking demand refers to the specific demand parameters parsed from the parking command, including scene type, reference target attributes, parking space physical characteristics, or user preferences. The parking space recommendation device receives and parses the parking command in real time to obtain the user's parking demand or the parking demand of the current scene.

[0087] Step S20: Based on the parking demand, perform multi-dimensional data matching to obtain parking space recommendation results, wherein the multi-dimensional data includes multiple items from static map data, user preference data, and real-time dynamic data.

[0088] It is understandable that the parking space recommendation result refers to one or more recommended parking space sequences, sorted in descending order of matching degree, output after calculation. The parking space recommendation device performs multi-dimensional data matching between the parsed parking demand and static map data, user preference data, and real-time dynamic data to obtain the parking space recommendation result.

[0089] In one possible implementation, the following steps are included prior to step S10:

[0090] When the vehicle is in a preset parking environment, based on the scene accuracy of the preset parking environment, static map data and user preference data corresponding to the scene accuracy are loaded, and real-time dynamic data is collected. The static map data includes at least one of real-time modeling map and structured map, and the user preference data includes at least one of general historical preference data and specific garage historical data.

[0091] It should be noted that the structured map refers to a pre-stored, high-precision structured map of the current parking garage. The real-time modeling map refers to a preliminary map constructed collaboratively by the device through 360° visual scanning using surround-view cameras and 3D radar ranging, containing the basic parking space outline and the location of core facilities. General historical preference data refers to a common requirement framework applicable to all parking garages, covering users' general preferences for parking space physical attributes (such as width and lighting), relative location (such as proximity to entrances / exits), and parking difficulty. Specific parking garage historical data refers to a unique behavioral profile strongly bound to the current parking garage, primarily including the user's historical parking records in that garage, scenario-based selection rules (such as "fixed selection of parking space A in shopping mall B1 parking garage - shopping scenario"), and parking space evaluation records. Preset parking environment refers to the preset environment for intelligent parking recommendations, typically various parking lots or garages. Scene accuracy refers to the parking recommendation device's familiarity with the current preset parking environment and the level of detail in the data it possesses, used to determine the device's strategy for accessing data resources. The parking space recommendation device loads corresponding static map data and user preference data based on its familiarity with the current preset parking environment and the level of detail of the data it possesses, and also collects real-time dynamic data.

[0092] Furthermore, referring to Figure 2 , Figure 2 A flowchart illustrating the intelligent parking space recommendation method is provided. When a vehicle enters the garage, the parking space recommendation device activates the parking space memory module and initiates an initialization process, simultaneously loading and activating three types of core data, such as... Figure 2 Step S1 is shown in the following:

[0093] Loading static map data:

[0094] Specifically, if the parking space recommendation device's parking space memory module has a pre-stored high-precision structured map of the current parking garage, then that high-precision structured map will be invoked. This map specifically covers the following:

[0095] Full-dimensional attribute information of parking spaces and fixed facilities: At the parking space level, key parameters include number, area (such as zone A / B / C / D), size type (standard parking space / wide parking space / micro parking space), location relationship (such as "parking space A12 is adjacent to elevator entrance E03"), and functional attributes (whether it is a charging parking space).

[0096] At the fixed facility level, static characteristics such as the type, color, and precise coordinates of facilities such as columns, fire hydrants, stairwells, charging piles, and entrances / exits are recorded.

[0097] Specifically, if the parking space recommendation device's parking space memory module does not have a pre-existing high-precision structured map of the current parking garage, the parking space recommendation device enters a real-time modeling mode. Through the coordinated work of 360° visual scanning by surround-view cameras and 3D ranging by radar, it quickly constructs a real-time modeling map containing the basic parking space outline and the location of core facilities, ensuring that scene perception can still be achieved even without pre-stored data.

[0098] Furthermore, even if the parking space memory module has a high-precision structured map of the current garage pre-stored, the parking space recommendation device can enter the real-time modeling mode to generate a real-time modeling map, which is used to update the information and visually compensate the high-precision structured map.

[0099] Accessing user preference data:

[0100] Specifically, if the parking space recommendation device's parking space memory module has a pre-stored exclusive behavior profile bound to the current parking space, i.e., specific parking space historical data, then the specific parking space historical data will be retrieved. This data specifically includes:

[0101] Garage semantic identifiers: such as "Recommended parking space in shopping mall B1 garage", "Recommended parking space in residential community underground second floor garage", and historical parking records (including parking space number, area, parking time, and automatic parking success rate).

[0102] Core Scenarios: Based on the user's activity purpose, core scenarios such as "shopping / charging / working / visiting relatives" are identified. These core scenarios are obtained by associating with navigation destinations, such as triggering a shopping scenario when navigating to a shopping mall or supermarket, or by being actively marked by the user, such as manually selecting a charging scenario.

[0103] Fixed selection rules for each scenario: such as "Parking space recommendation: shopping mall B1 garage - shopping scenario → select only parking space in area A (near supermarket entrance)" and "Parking space recommendation: shopping mall B1 garage - charging scenario → always select parking space C08 (next to charging pile P02)");

[0104] Parking space evaluation records: Through two channels, namely "user proactive feedback" (such as voice complaints that "the parking space is too narrow") and "system behavior inference" (such as rejecting parking space B03 three times in a row), positive and negative labels such as "recommended / not recommended" are added to the parking space to ensure the objectivity of the evaluation.

[0105] Specifically, the parking space recommendation device accesses general user historical preference data when a vehicle enters the garage. This data includes:

[0106] General attribute preferences: Fixed requirements for physical attributes such as parking space width, lighting, spatial permeability, water and oil stains on the ground, and the presence of charging piles;

[0107] General location preference: A consistent tendency toward the relative position of parking spaces and fixed facilities, such as prioritizing proximity to entrances and exits and avoiding parking spaces next to sidewalks;

[0108] General parking difficulty preference: a risk threshold set based on the capabilities of the automated parking system.

[0109] Dynamic information is updated in real time:

[0110] Specifically, when a vehicle enters the garage, the parking space recommendation system dynamically updates two types of key information based on real-time sensing by onboard sensors such as radar and cameras:

[0111] Status information: Focus on verifying and updating easily changeable content in the static map, including the real-time status of parking spaces, such as switching from "empty" to "occupied" or vice versa, and the status of the parking space ground, such as adding attributes that are inconsistent with the static data, such as water accumulation / oil stains.

[0112] Instance-based information: Accurately identify and record dynamic targets within the garage, covering vehicles and their attributes, such as real-time location, brand, color, and model class; personnel and their attributes, such as gender, age, clothing, and behavioral status, such as walking / standing; temporary facilities and their attributes, such as the type, size, and movement trajectory of cleaning equipment like floor scrubbers, construction tools like warning cones, and temporary storage items.

[0113] In this implementation, data is layered and adapted, that is, static maps distinguish between pre-stored high-precision maps and real-time modeling of unknown garages, and user preferences distinguish between specific garage-specific files and general common needs. This covers three core data categories: fixed facilities, user habits, and real-time dynamics, and can accurately match data dimensions according to the scenario, laying a comprehensive and accurate data foundation for recommendations.

[0114] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 Step S20, the parking space recommendation method further includes steps S01~S02:

[0115] Step S01: Determine the recommendation dimension corresponding to the parking demand, wherein the recommendation dimension includes any one of the basic recommendation dimension and the combined recommendation dimension;

[0116] Understandably, a basic recommendation dimension refers to the single semantic logic direction relied upon when performing parking space matching, including automatic scene matching, reference target-oriented filtering, and precise adaptation of parking space attributes. A combined recommendation dimension refers to a composite matching pattern that simultaneously includes two or more basic recommendation dimensions, used to handle complex parking needs involving multiple factors in user instructions. After parsing the parking need, the parking space recommendation device determines its corresponding recommendation dimension. For example, if the user instruction is simply "shopping," then it determines the basic recommendation dimension as automatic scene matching. If the instruction is "parking in a space without standing water and near the elevator while shopping," since it involves the scene (shopping), attribute (no standing water), and reference target (near the elevator), then the recommendation dimension corresponding to the parking need is determined to be a combined recommendation dimension. The combined pattern retrieval process is then initiated, the weights of multiple dimensions are fused and calculated, and finally, the parking space recommendation result with the highest overall matching degree is output.

[0117] Step S02: Select the target parking space node corresponding to the recommendation dimension from the parking space nodes in the preset semantic graph, and determine the parking space recommendation result based on the target parking space node.

[0118] It should be noted that the target parking space node refers to the set of candidate parking space nodes that meet the parking requirements identified in the semantic graph. The parking space recommendation device filters the parking space nodes in the preset semantic graph to obtain the set of candidate parking space nodes that meet the parking requirements corresponding to the current recommendation dimension, and outputs the final parking space recommendation result based on the set of candidate parking space nodes.

[0119] In one possible implementation, the following steps are included prior to step S20:

[0120] Acquire multi-dimensional data and convert the multi-dimensional data into semantic graph nodes, wherein the semantic graph nodes include scene nodes, reference target nodes, parking space attribute nodes, user preference nodes, and parking space nodes;

[0121] Understandably, parking space recommendation devices abstract multi-dimensional data in a structured way according to their semantic categories, forming entities that can be related to each other in a graph.

[0122] For example, such as Figure 2 As shown, after acquiring multi-dimensional data in step S1, the device converts the multi-dimensional data into semantic graph nodes in step S2.

[0123] The association weights between nodes in the semantic graph are quantified, and node connections between the semantic graph nodes are established based on the association weights to obtain a preset semantic graph.

[0124] It should be noted that the parking space recommendation device calculates the relationship strength, i.e., the association weight, between different nodes, and creates or updates the connection relationship between nodes in the edge layer of the semantic graph based on the calculated association weight, thereby forming a network structure with quantified relationship strength, i.e., the preset semantic graph.

[0125] For example, such as Figure 2 As shown in step S2, after constructing the node layer, the device quantifies various association weights. For example, for a shopping scenario node and a specific parking space node, the association weight is calculated by combining the distance weight from the parking space to the mall entrance and the historical frequency weight of the user selecting the parking space in this scenario. Similarly, for a male reference target node, the association weight with the parking space is calculated using different distance threshold formulas based on the type of instruction to approach or avoid. All calculated weights are used to establish node connections in the graph, ultimately obtaining a complete and usable preset semantic graph, which supports the recommendation in step S3.

[0126] In one feasible implementation, the step of quantifying the association weights between nodes in the semantic graph includes:

[0127] Calculate the straight-line distance from each parking space corresponding to the parking space node to the key facility corresponding to the scene node;

[0128] Understandably, scene nodes are generated based on navigation destination associations and user-generated annotations, including sub-nodes such as shopping, charging, working, and visiting relatives. Each sub-node is associated with a corresponding location requirement, such as shopping being associated with the entrance / exit of a shopping mall. Key facilities refer to fixed facilities strongly correlated with specific scene nodes and capable of defining the core location requirements of that scene. For example, the key facility in a shopping scene is the mall entrance / exit, in a charging scene it is a charging station, and in an working scene it is the office building elevator lobby. The parking space recommendation device calculates the shortest spatial distance between the center point of the parking space and the coordinates of the key facility using Euclidean geometry algorithms.

[0129] The number of times each parking space is entered in the scene corresponding to each scene node is counted;

[0130] It should be noted that the parking count refers to the total number of times a user successfully parks their vehicle in a specific parking space when the vehicle is in a particular scenario, based on historical data records. The parking space recommendation device quantifies the frequency of a user's historical choices in a specific scenario, transforming implicit personal habits into calculable preference weights.

[0131] Based on the straight-line distance and the number of parking entries, the scene association weight between the parking space node and the scene node in the semantic graph is quantified.

[0132] Understandably, scene association weights represent the degree of matching between a parking space and a user's needs in a specific scene. The parking space recommendation device calculates the matching degree between parking space nodes and scene nodes by fusing straight-line distance and parking frequency.

[0133] Specifically, scene-related weights:

[0134]

[0135] in, Parking space To the scene The straight-line distance to the corresponding key facilities; The maximum distance from a parking space in the garage to this critical facility; Historical scenes Select parking space below The number of times; Historical scenes Total number of parking times; Distance weighting coefficient.

[0136] In one feasible implementation, the step of quantifying the association weights between nodes in the semantic graph further includes:

[0137] Calculate the straight-line distance from each parking space corresponding to the parking space node to each reference target of the reference target node;

[0138] It should be noted that the reference target node refers to the node in the semantic graph that represents a dynamic or static reference object, including sub-nodes such as specific people, specific vehicles, sidewalks, and temporary facilities generated based on instance class information. The parking space recommendation device calculates the Euclidean distance between the coordinates of the parking space center point and the real-time coordinates of the reference target.

[0139] Based on the straight-line distance, the target association weights between the parking space node and the scene node in the semantic graph are quantified for different semantic types, wherein the semantic types include tendency type and avoidance type.

[0140] Understandably, semantic type refers to the user's stance towards the reference target in their instructions. A preference type indicates the user wants to get closer to the target, while an avoidance type indicates the user wants to stay away. The parking space recommendation device quantifies the target association weights for different semantic types based on straight-line distance. By designing differentiated weight calculation functions for these two opposing semantic intentions, the same spatial distance data can generate correct recommendation logic based on the user's drastically different needs.

[0141] Specifically, target association weights:

[0142] For a specific target (e.g., stopping next to a man):

[0143]

[0144] When the instruction is to avoid a target (e.g., move away from the sidewalk):

[0145]

[0146] in, Parking space To the reference target The straight-line distance; : Effective distance threshold for approaching the target (e.g., 3 meters, based on the accuracy of the vehicle-mounted sensor). : Safe distance threshold for avoiding targets (e.g., 5 meters, set based on user preferences and garage specifications).

[0147] In one feasible implementation, the step of quantifying the association weights between nodes in the semantic graph further includes:

[0148] Based on the multi-dimensional data, the matching degree between each parking space corresponding to the parking space node and the parking space attribute corresponding to the parking space attribute node is quantified.

[0149] It should be noted that parking space attribute nodes refer to nodes in the semantic graph that represent the inherent physical characteristics or real-time status of parking spaces, such as sub-nodes for size type, functional attributes, ground condition, and lighting conditions. Matching degree is used to quantify the degree to which a specific parking space node conforms to a particular attribute. The parking space recommendation device calculates the matching degree between parking space attributes and demand. For example, when the parking space attribute node is "no standing water," the device detects the ground condition of parking space P based on real-time dynamic data: if there is no standing water, the matching degree between parking space P and this attribute is 1; if there is standing water, the matching degree is 0.

[0150] Based on the matching degree, the attribute association weight between the parking space node and the parking space attribute node in the semantic graph is quantified.

[0151] Understandably, attribute association weights are used to characterize the degree to which a parking space node meets user needs across specific attribute dimensions. The parking space recommendation device quantifies the attribute association weights between parking space nodes and parking space attribute nodes through matching degrees.

[0152] Specifically, attribute association weights:

[0153]

[0154] in, Parking space For attributes The satisfaction level is binarized to a value of 0 or 1 (e.g., 1 if there is no water accumulation, 0 if there is water accumulation).

[0155] In one feasible implementation, the step of quantifying the association weights between nodes in the semantic graph further includes:

[0156] Based on the number of positive and negative feedbacks from the user for each parking space corresponding to the parking space node, a specific preference matching score is calculated.

[0157] It's important to note that positive feedback counts refer to the cumulative number of positive evaluations from users through active selection, positive reviews, or system-inferred approval behaviors (such as not leaving immediately after successfully parking). Negative feedback counts refer to the cumulative number of negative evaluations from users through active negative reviews, avoidance selection, or system-inferred rejection behaviors (such as repeatedly refusing a parking space or briefly changing parking spaces due to parking issues). The specific preference matching score is a quantitative score that integrates both positive and negative feedback, used to dynamically measure how well a parking space matches a user's specific preferences. The parking space recommendation device obtains the specific preference matching score by simultaneously counting and integrating both positive and negative feedback behaviors.

[0158] Based on the preference weights of the general preference data in the multi-dimensional data, the general preference matching score of each parking space that satisfies the general preference is calculated;

[0159] Understandably, the general preference matching score is a comprehensive quantitative value used to globally assess the degree to which a parking space meets the common needs of users across different parking garages. Preference weights are dynamic coefficients reflecting the degree of importance users place on various general preferences. The parking space recommendation device calculates a comprehensive general preference matching score for each parking space based on the preference weights in the general preference data.

[0160] Based on the specific preference matching score and the general preference matching score, the preference association weight between the parking space node and the user preference node in the semantic graph is quantified.

[0161] It should be noted that the preference association weight represents the degree to which a parking space node matches the user's overall preferences. The parking space recommendation device weights and combines specific preference matching scores and general preference matching scores to obtain the preference association weight between parking space nodes and user preference nodes.

[0162] Specifically, preference association weights:

[0163]

[0164] in: It is the preference fusion coefficient, with a value range of [0,1], used to adjust the influence ratio of the two types of preferences; Historically, choosing a parking space The number of times; Historically, users have had a positive attitude towards parking spaces. The number of positive reviews; Historically, users have had a positive attitude towards parking spaces. The number of negative reviews; This represents the total number of times the car has been parked in the garage in history. It is the weight of the i-th general preference; It is a parking space The degree of satisfaction with the i-th general preference.

[0165] In this implementation, scenarios, reference targets, parking space attributes, and user preferences are transformed into semantic graph nodes. By quantifying association weights, such as scenario-parking space distance weights and target-parking space avoidance weights, node connections are established, making unstructured requirements computable.

[0166] Based on the first embodiment of this application, in the third embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 4 Step S02, the parking space recommendation method further includes steps A01~A03:

[0167] Step A01: Filter and correct the association weights of parking space nodes in the preset semantic graph based on the recommendation dimension to obtain candidate parking space nodes;

[0168] Understandably, after filtering parking space nodes in the preset semantic graph through recommendation dimensions, the parking space recommendation device will also adjust the association weight of the filtered nodes to obtain candidate parking space nodes. Through the two-step processing mechanism of filtering and then adjusting, it ensures that the candidate parking space set is highly relevant to the user's core needs, and also incorporates personalized preference fine-tuning on this basis.

[0169] Step A02: Select target parking space nodes with a parking difficulty coefficient less than the first threshold from the candidate parking space nodes;

[0170] It should be noted that the parking difficulty coefficient is used to characterize the operational complexity and risk level of automatically parking a vehicle into a specific parking space. This coefficient is based on the capability settings of the Automated Parking System (APS). The parking space recommendation device uses a first threshold... By filtering out target parking space nodes that meet the requirements for automatic parking of vehicles, the practicality of the recommendations is improved.

[0171] Step A03: Sort and output the target parking space nodes based on the corrected association weights to obtain the parking space recommendation results.

[0172] Understandably, the parking space recommendation device sorts the target parking space nodes in descending order according to their corresponding corrected association weights, and the final ordered list is the parking space recommendation result.

[0173] Furthermore, the device will output the sorted list as the final parking space recommendation result to the human-computer interaction interface for users to view and select, and record the user's selection behavior.

[0174] In one feasible implementation, step A01 includes:

[0175] The scene nodes corresponding to the scene are automatically matched and recommended from the nodes of the preset semantic graph;

[0176] It should be noted that the parking space recommendation device automatically identifies and activates a specific scene node in the semantic graph by automatically matching and recommending corresponding nodes from the preset semantic graph. This scene node is determined based on demand information.

[0177] Traverse the first association edge between the scene node and the parking space node, and filter out candidate parking space nodes whose scene association weight of the first association edge is greater than or equal to the second threshold.

[0178] Understandably, the second threshold is a scene adaptation threshold, such as 0.6. The parking space recommendation device traverses the scene-parking space association edges in the semantic association graph and filters out a set of candidate parking spaces whose association weight with the current scene node is greater than or equal to the second threshold.

[0179] Based on the preference association weights corresponding to the scene preference nodes in the user preference nodes, the scene association weights of the candidate parking space nodes are corrected to obtain the corrected candidate parking space nodes.

[0180] It should be noted that the scene preference node is a specific sub-node of the user preference node. It records the user's additional preference tendencies for certain parking spaces or areas in a specific scene. The parking space recommendation device integrates the preference association weights carried by the scene preference node into the existing scene association weights of the candidate parking space node in a weighted summation manner to obtain the corrected candidate parking space node.

[0181] Specifically, when the device identifies a scene node through navigation commands, it will search for parking space nodes based on the scene node. The search process is as follows:

[0182] 1. Candidate Filtering: Filtering based on the scenario-parking space association edge. The candidate set is obtained from parking spaces with a scene adaptation threshold of 0.6. ;

[0183] 2. Weight Fusion: The system fuses scenario-related weights and preference-related weights to calculate the overall parking space score.

[0184]

[0185] 3. Difficulty Filtering and Sorting: Eliminating docking difficulty coefficients (e.g., parking space 3) is calculated according to... Output in descending order.

[0186] In one feasible implementation, step A01 further includes:

[0187] The reference target node corresponding to the reference target-oriented recommendation is determined from the nodes of the preset semantic graph;

[0188] It is understandable that the parking space recommendation device matches and locks one or more specific child nodes in the reference target node layer of the semantic graph by matching the dynamic or static targets described in the user's natural language instructions. That is, it recommends the reference target node corresponding to the reference target.

[0189] Traverse the second association edge between the reference target node and the parking space node to filter out candidate parking space nodes that meet the distance requirements;

[0190] It should be noted that after determining the reference target node, the parking space recommendation device traverses all the associated edges connected to that node and, based on the semantic type of the user's instruction (preference or avoidance), determines whether the straight-line distance between the parking space node and the reference target node meets the preset distance threshold condition.

[0191] Based on the preference association weight of the location avoidance preference node in the user preference node, the target association weight of the candidate parking space node is corrected to obtain the corrected candidate parking space node.

[0192] Understandably, the location avoidance preference node is a specific sub-node within the user preference node. It records the user's tendency to avoid parking spaces relative to certain types of reference targets, such as avoiding parking spaces next to temporary facilities or sidewalks. When a candidate parking space node has a strong spatial association with a reference target node that needs to be avoided, the parking space recommendation device reduces or downweights the target association weight based on the weight of the location avoidance preference node, resulting in a corrected candidate parking space node.

[0193] Specifically, when the device receives a reference target instruction, it will search for parking space nodes based on the reference target node. The search process is as follows:

[0194] 1. Target matching and candidate filtering: Based on the target type parsing of instructions, match reference target nodes. ,pass Filter parking spaces that meet the distance requirements to obtain a candidate set. ;

[0195] 2. Weight Fusion: The target association weight and location preference weight are fused to calculate the overall score.

[0196]

[0197] 3. Difficulty Filtering and Sorting: Eliminating Parking spaces, according to Output in descending order.

[0198] In one feasible implementation, step A01 further includes:

[0199] The parking space attribute nodes corresponding to the parking space attribute adaptation recommendation are determined from the nodes of the preset semantic graph;

[0200] It should be noted that, based on the parsing results of user commands, the parking space recommendation device matches and locks one or more specific child nodes in the parking space attribute node layer of the semantic graph, such as ground status nodes, size type nodes, lighting condition nodes, and functional attribute nodes.

[0201] Traverse the third association edge between the parking space attribute node and the parking space node, and filter out candidate parking space nodes whose attribute association weight of the third association edge is greater than the third threshold.

[0202] Understandably, the third threshold is a preset attribute satisfaction threshold used to determine whether the parking space meets the minimum acceptable standard for satisfying user needs in a specific attribute dimension. The parking space recommendation device filters parking space nodes by setting attribute association weight thresholds, ensuring that selected candidate parking spaces have a high degree of matching in the key attributes specified by the user. This effectively filters out parking spaces that perform well in other aspects but fail to meet the user's core attribute needs, allowing the recommendation results to accurately respond to the user's specific requirements for the physical or state characteristics of the parking space.

[0203] Based on the preference association weights of the attribute demand preference nodes in the user preference nodes, the attribute association weights of the candidate parking space nodes are corrected to obtain the corrected candidate parking space nodes.

[0204] It should be noted that the attribute demand preference node is a specific sub-node within the user preference node. It records the user's preference for specific physical attributes of parking spaces (e.g., wide-body parking spaces, charging parking spaces, no standing water). The parking space recommendation device integrates the preference association weights carried by the attribute demand preference node into the basic attribute association weights calculated based on the current state of the candidate parking space node in a weighted summation manner. This process corrects the candidate parking space node, resulting in a corrected candidate parking space node.

[0205] Specifically, when the device receives a user's instruction regarding parking space attributes, it will search for parking space nodes based on the parking space attribute nodes. The search process is as follows:

[0206] 1. Attribute Matching and Candidate Filtering: Based on the attribute requirements parsed from the instruction, match parking space attribute nodes. ,pass filter Parking spaces that meet the attribute threshold (e.g., 0.8) are selected as the candidate set. ;

[0207] 2. Weight Fusion: Combine attribute association weights and attribute preference weights to calculate the overall score.

[0208]

[0209] 3. Difficulty Filtering and Sorting: Eliminating Parking spaces, according to Output in descending order.

[0210] In one feasible implementation, step A01 further includes:

[0211] From the nodes of the preset semantic graph, determine the single-dimensional nodes corresponding to the single recommendation dimension in the combined recommendation dimensions;

[0212] It's understandable that the combined recommendation dimensions include multiple aspects of automatic scenario matching recommendation, reference target-oriented recommendation, and parking space attribute adaptation recommendation. A single-dimensional node, on the other hand, refers to a node entity in the semantic graph that precisely corresponds to each individual dimensional requirement contained in the current user's complex command. The parking space recommendation device accurately breaks down complex combined commands and maps them to independent single-dimensional nodes in the semantic graph.

[0213] The fourth association edge between the single-dimensional node and the parking space node is traversed respectively. Based on the semantic type in the parking requirement, candidate parking space nodes that meet the parking requirement are selected.

[0214] It should be noted that the parking space recommendation device sequentially visits the edges connecting each determined single-dimensional node to all parking space nodes. During the traversal and filtering process, it uses the corresponding judgment logic that conforms to the semantic type for different single-dimensional nodes to obtain candidate parking space nodes that meet the parking requirements.

[0215] Based on the association weights of each node in the user preference node, the association weights of candidate parking space nodes in different dimensions are corrected, and the corrected association weights are summed in multiple dimensions to obtain the candidate parking space nodes after weighted summation.

[0216] It is understandable that each node in the user preference node refers to a preference sub-node related to each individual dimension in the current combined recommendation dimension. In the combined mode, the parking space recommendation device adjusts the preference weights of each dimension, and then uses preset dimension weight coefficients to linearly weight and fuse the scene association weights, target association weights, and attribute association weights after their respective dimension preference adjustments to obtain a weighted sum of candidate parking space nodes.

[0217] Specifically, when a user command involves multiple dimensions simultaneously, a combined recommendation dimension is triggered, as shown below:

[0218] After multi-dimensional screening, a comprehensive score is calculated, such as through a "multi-weighted summation" method:

[0219]

[0220] in, , , Weighting coefficients for scenario, attribute, and reference target dimensions.

[0221] In this implementation, three basic modes plus one combined mode cover almost all parking scenarios: when the user has no explicit instructions, scene matching can be automatically triggered via navigation; when there are specific needs, filtering can be based on reference targets; when there are attribute requirements, precise matching of parking space physical characteristics can be achieved; and when multiple needs overlap, combined recommendations can be achieved through weighted fusion. This hierarchical and categorized recommendation logic allows users with different needs to obtain suitable parking spaces, greatly improving the flexibility of use.

[0222] Based on the first embodiment of this application, in the fourth embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description and will not be repeated hereafter. Furthermore, before step S10, the parking space recommendation method further includes step B01 or step B02:

[0223] Step B01: When the user's voice information is received, the voice information is parsed into semantic demand to obtain demand information, and a parking instruction is generated based on the demand information. The demand information includes at least scene information, attribute information and target information.

[0224] It's important to note that contextual information refers to the surrounding information related to the parking purpose identified from the voice, such as shopping or charging. Attribute information refers to the user's requirements for the physical or state characteristics of the parking space, such as no standing water or shade. Target information refers to dynamic or static objects mentioned in the user's instructions as spatial references, such as next to a man or near an elevator. When the parking space recommendation device receives the user's voice information, it uses natural language processing technology to parse the user's raw, unstructured voice to obtain structured data, i.e., the demand information.

[0225] or,

[0226] Step B02: When the user does not issue a parking instruction, the system automatically determines the demand information, including the scene information, based on the navigation information, and generates the parking instruction based on the demand information.

[0227] Understandably, the parking space recommendation device intelligently associates the navigation destination with the parking scenario. When the user does not issue a parking instruction, if the device detects that the destination of the navigation information is "parking space recommendation mall", it will automatically associate the destination with the shopping scenario, thereby determining the demand information containing the shopping scenario information, and generating a parking instruction that automatically matches and recommends the trigger scenario based on this demand information, thus achieving seamless triggering.

[0228] In this implementation, natural interaction is achieved through semantic parsing of needs and seamless triggering: On the one hand, users can express their needs using everyday language commands (such as "Don't park in a flooded parking space" or "Park near the elevator when shopping"). The system can automatically parse key information such as the scenario, attributes, and goals in the commands without requiring users to manually select category tags. On the other hand, when users do not actively give commands, the system can automatically trigger recommendations based on navigation destinations (such as navigating to a shopping mall to trigger a "shopping" scenario) and historical behavior, reducing the need for users to actively perform operations. This interaction mode is more in line with users' daily expression habits, significantly lowers the operational threshold, and is especially user-friendly for middle-aged and elderly users and drivers.

[0229] Based on the first embodiment of this application, in the fifth embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. On this basis, refer to Figure 5 After step S20, the parking space recommendation method further includes steps C01 to C04:

[0230] Step C01: Calculate the number of times the data attributes of the vehicle parking process overlap with the data attributes of the general historical preference data, wherein the behavioral data includes at least one of active selection data, active feedback data, and system inference data;

[0231] It should be noted that "active selection data" refers to the data generated and recorded when a user explicitly selects a specific parking space from the system's candidate parking space list through touch, voice confirmation, or other means. This data directly reflects the user's decision. "Active feedback data" refers to the data generated when a user, without actively selecting a parking space, actively provides positive or negative feedback on a parking space or its attributes through voice evaluation (e.g., "This parking space is too narrow"), rating interfaces, or other channels. "System inference data" refers to data that indirectly reflects user preferences, inferred by the system based on a series of indirect user actions (e.g., repeatedly rejecting system-recommended parking spaces of the same type, or leaving a parking space very quickly after successfully parking) using logical rules or models. "Data attributes" refers to the parking space characteristic dimensions jointly described by behavioral data and general historical preference data. "Overlap count" refers to the cumulative number of times, within a single parking event or period, the characteristic selection reflected in the user's behavioral data aligns with the characteristic requirements described by a certain general historical preference data point in terms of data attributes. The parking space recommendation device quantifies the matching frequency between users' immediate behavior and long-term general preferences, obtaining the number of overlaps.

[0232] Step C02: Based on the data type of the general historical preference data, classify the number of overlaps to obtain the target number of overlaps corresponding to different data types;

[0233] Understandably, the parking space recommendation device performs refined classification and statistics on the original number of overlaps according to preference type to obtain the target overlap number for different data types.

[0234] Step C03: Based on the number of target overlaps and the differential weights corresponding to different data types, calculate the preference scores of the general preference data for different data types.

[0235] It should be noted that differentiated weights refer to different weight coefficients assigned to different types of behavioral data. The parking space recommendation device obtains preference scores for different types of general preference data by introducing differentiated weights and weighting the number of times the target overlaps.

[0236] Step C04: Based on the preference score, calculate the updated dynamic weights of the general preference data for different data types, wherein the dynamic weights are used to update the preference weights of the general preference data.

[0237] Understandably, the parking space recommendation device converts preference scores into standardized dynamic weights. Based on these dynamic weights, the general preference weights can be updated, continuously adapting to user habits, avoiding rigid recommendations, and resulting in a better long-term user experience.

[0238] In this embodiment, the established preference update closed loop collects behavioral data through multiple channels, including active selection, active feedback, and system inference. This captures both explicitly expressed user needs and implicit habits. At the same time, through differentiated weight calculation and normalization processing, preference weights are dynamically adjusted according to behavior, avoiding excessive influence of a single operation on the recommendation results.

[0239] Furthermore, all user-related data involved in this application (such as user attribute data, user behavior data, and user navigation data) were obtained with the user's permission or consent; that is, when this application is applied to a specific product or technology, user permission is required to obtain and process the relevant data, and the processing of the relevant data must comply with the relevant laws, regulations, and regulatory standards of the relevant countries and regions.

[0240] For example, when it is necessary to obtain a user's behavioral data, a prompt to obtain behavioral data can be displayed on the user's terminal. After receiving confirmation from the user regarding the prompt, the terminal can obtain the user's behavioral data. (See below.) Figure 6 As shown:

[0241] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the parking space recommendation method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0242] This application also provides a parking space recommendation device, please refer to... Figure 7 The parking space recommendation device includes:

[0243] The response module 10 is used to receive parking instructions and determine parking requirements based on the parking instructions;

[0244] The matching module 20 is used to perform multi-dimensional data matching based on the parking demand to obtain parking space recommendation results, wherein the multi-dimensional data includes multiple items from static map data, user preference data, and real-time dynamic data.

[0245] Optionally, the response module includes:

[0246] The loading submodule is used to load static map data and user preference data corresponding to the scene accuracy based on the scene accuracy of the preset parking environment when the vehicle is in a preset parking environment, and to collect real-time dynamic data. The static map data includes at least one of real-time modeled map and structured map, and the user preference data includes at least one of general historical preference data and specific garage historical data.

[0247] A generation submodule is used to perform semantic parsing of the voice information received from a user to obtain demand information, and generate a parking instruction based on the demand information, wherein the demand information includes at least scene information, attribute information, and target information; or, when the user does not issue a parking instruction, automatically determine the demand information including the scene information based on navigation information, and generate the parking instruction based on the demand information.

[0248] The calculation submodule is used to calculate the number of times the data attributes of the vehicle parking process's behavioral data overlap with the data attributes of the general historical preference data, wherein the behavioral data includes at least one of active selection data, active feedback data, and system inference data; based on the data type of the general historical preference data, the overlap number is classified to obtain the target overlap number corresponding to different data types; based on the target overlap number and the differential weights corresponding to different data types, the preference score of the general preference data for different data types is calculated; based on the preference score, the updated dynamic weight of the general preference data for different data types is calculated, wherein the dynamic weight is used to update the preference weight of the general preference data.

[0249] Optionally, the matching module includes:

[0250] The filtering submodule is used to determine the recommendation dimension corresponding to the parking demand, wherein the recommendation dimension includes any one of the basic recommendation dimension and the combined recommendation dimension; from the parking space nodes of the preset semantic graph, the target parking space node corresponding to the recommendation dimension is obtained, and the parking space recommendation result is determined based on the target parking space node.

[0251] Optionally, the filtering submodule includes:

[0252] A unit is established to acquire multi-dimensional data and convert the multi-dimensional data into semantic graph nodes, wherein the semantic graph nodes include scene nodes, reference target nodes, parking space attribute nodes, user preference nodes, and parking space nodes; the association weights between nodes in the semantic graph nodes are quantified, and node connections between the semantic graph nodes are established based on the association weights to obtain a preset semantic graph.

[0253] Optionally, the establishment unit includes:

[0254] The first quantization subunit is used to calculate the straight-line distance between each parking space corresponding to the parking space node and the key facility corresponding to the scene node; count the number of times each parking space is parked in the scene corresponding to each scene node; and quantify the scene association weight between the parking space node and the scene node in the semantic graph node based on the straight-line distance and the number of parking entries.

[0255] The second quantization subunit is used to calculate the straight-line distance between each parking space corresponding to the parking space node and each reference target of the reference target node; based on the straight-line distance, quantize the target association weights between the parking space node and the scene node in the semantic graph for different semantic types, wherein the semantic types include tendency type and avoidance type.

[0256] The third quantization subunit is used to quantify the matching degree between each parking space corresponding to the parking space node and the parking space attribute corresponding to the parking space attribute node based on the multi-dimensional data; and to quantify the attribute association weight between the parking space node and the parking space attribute node in the semantic graph node based on the matching degree.

[0257] The fourth quantization subunit is used to calculate a specific preference matching score based on the number of positive and negative feedbacks from the user for each parking space corresponding to the parking space node; calculate a general preference matching score for each parking space that satisfies the general preference based on the preference weights of the general preference data in the multi-dimensional data; and quantify the preference association weights between the parking space node and the user preference node in the semantic graph node based on the specific preference matching score and the general preference matching score.

[0258] Optionally, the filtering submodule includes:

[0259] The output unit is used to filter parking space nodes in the preset semantic graph based on the recommendation dimension and correct the association weight to obtain candidate parking space nodes; filter target parking space nodes with a parking difficulty coefficient less than a first threshold from the candidate parking space nodes; and sort and output the target parking space nodes based on the corrected association weight to obtain the parking space recommendation result.

[0260] Optionally, the output unit includes:

[0261] The first correction subunit is used to determine the scene node corresponding to the automatic matching recommendation of the scene from the nodes of the preset semantic graph; traverse the first association edge between the scene node and the parking space node, and filter out the candidate parking space nodes whose scene association weight of the first association edge is greater than or equal to a second threshold; and correct the scene association weight of the candidate parking space node based on the preference association weight corresponding to the scene preference node in the user preference node to obtain the corrected candidate parking space node.

[0262] The second correction subunit is used to determine the reference target node corresponding to the reference target orientation recommendation from the nodes of the preset semantic graph; traverse the second association edge between the reference target node and the parking space node, and filter out candidate parking space nodes that meet the distance requirements; and correct the target association weight of the candidate parking space node based on the preference association weight of the location avoidance preference node in the user preference node to obtain the corrected candidate parking space node.

[0263] The third correction subunit is used to determine the parking space attribute node corresponding to the parking space attribute adaptation recommendation from the nodes of the preset semantic graph; traverse the third association edge between the parking space attribute node and the parking space node, and filter out candidate parking space nodes whose attribute association weight of the third association edge is greater than a third threshold; and correct the attribute association weight of the candidate parking space node based on the preference association weight of the attribute demand preference node in the user preference node to obtain the corrected candidate parking space node.

[0264] The fourth correction subunit is used to determine the single-dimensional nodes corresponding to the single recommendation dimension in the combined recommendation dimension from the nodes of the preset semantic graph; traverse the fourth association edge between the single-dimensional node and the parking space node respectively, and filter out candidate parking space nodes that meet the parking requirements based on the semantic type in the parking requirements; based on the association weight of each node in the user preference nodes, correct the association weight of the candidate parking space nodes in different dimensions respectively, and perform multi-dimensional weighted summation on the corrected association weights to obtain the candidate parking space nodes after weighted summation.

[0265] The parking space recommendation device provided in this application, employing the parking space recommendation method in the above embodiments, can solve the technical problem of parking space recommendation. Compared with the prior art, the beneficial effects of the parking space recommendation device provided in this application are the same as those of the parking space recommendation method provided in the above embodiments, and other technical features in the parking space recommendation device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0266] This application provides a parking space recommendation device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the parking space recommendation method in the above embodiment 1.

[0267] The following is for reference. Figure 8 The diagram illustrates a structural schematic suitable for implementing the parking space recommendation device in the embodiments of this application. The parking space recommendation device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, tablets, digital broadcast receivers, PDAs (Personal Digital Assistants), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 8 The parking space recommendation device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0268] like Figure 8 As shown, the parking space recommendation device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the parking space recommendation device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows the parking space recommendation device to communicate wirelessly or wiredly with other devices to exchange data. Although the figures show parking space recommendation devices with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented alternatively.

[0269] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0270] The parking space recommendation device provided in this application, employing the parking space recommendation method described in the above embodiments, can solve the technical problem of parking space recommendation. Compared with the prior art, the beneficial effects of the parking space recommendation device provided in this application are the same as those of the parking space recommendation method provided in the above embodiments, and other technical features of this parking space recommendation device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0271] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0272] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0273] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the parking space recommendation method in the above embodiments.

[0274] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0275] The aforementioned computer-readable storage medium may be included in the parking space recommendation device; or it may exist independently and not be installed in the parking space recommendation device.

[0276] The aforementioned computer-readable storage medium carries one or more programs that, when executed by the parking space recommendation device, cause the parking space recommendation device to: receive a parking instruction; determine parking demand based on the parking instruction; and perform multi-dimensional data matching based on the parking demand to obtain a parking space recommendation result, wherein the multi-dimensional data includes multiple items from static map data, user preference data, and real-time dynamic data.

[0277] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0278] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0279] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0280] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described parking space recommendation method, thereby solving the technical problem of parking space recommendation. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the parking space recommendation method provided in the above embodiments, and will not be repeated here.

[0281] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the parking space recommendation method described above.

[0282] The computer program product provided in this application can solve the technical problem of parking space recommendation. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the parking space recommendation method provided in the above embodiments, and will not be repeated here.

[0283] The above description is only a part of the embodiments of this application and does not limit the scope of protection of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the scope of protection of this application.

Claims

1. A parking space recommendation method, characterized in that, The parking space recommendation method includes: Receive parking instructions and determine parking requirements based on the parking instructions; Based on the parking demand, multi-dimensional data matching is performed to obtain parking space recommendation results. The multi-dimensional data includes multiple items from static map data, user preference data, and real-time dynamic data. The steps for performing multi-dimensional data matching based on the parking demand to obtain parking space recommendation results include: Determine the recommendation dimension corresponding to the parking demand, wherein the recommendation dimension includes any one of the basic recommendation dimension and the combined recommendation dimension; From the parking space nodes in the preset semantic graph, the target parking space nodes corresponding to the recommendation dimension are selected, and the parking space recommendation results are determined based on the target parking space nodes; The step of receiving a parking instruction and determining the parking requirement based on the parking instruction includes: The number of times the data attributes of the vehicle's behavioral data during the parking process overlap with the data attributes of the general preference data is calculated. The behavioral data includes at least one of active selection data, active feedback data, and system inference data. The general preference data is retrieved when the vehicle enters the garage. Based on the data type of the general preference data, the number of overlaps is classified to obtain the target number of overlaps corresponding to different data types; Based on the number of times the target overlaps and the differentiated weights corresponding to different data types, the preference scores of the general preference data for different data types are calculated. Based on the preference score, the updated dynamic weights of the general preference data for different data types are calculated. The dynamic weights are used to update the preference weights of the general preference data, and the preference weights of the general preference data are used to calculate the preference association weights between parking space nodes and user preference nodes in the preset semantic graph.

2. The parking space recommendation method as described in claim 1, characterized in that, Prior to the step of receiving a parking instruction and determining the parking requirement based on the parking instruction, the following steps are included: When the vehicle is in a preset parking environment, based on the scene accuracy of the preset parking environment, the static map data and user preference data corresponding to the scene accuracy are loaded, and real-time dynamic data are collected. The static map data includes at least one of a real-time modeled map and a structured map, and the user preference data includes at least one of general preference data and historical data of a specific parking garage.

3. The parking space recommendation method as described in claim 1, characterized in that, Before the step of performing multi-dimensional data matching based on the parking demand to obtain parking space recommendation results, the following steps are included: Acquire multi-dimensional data and convert the multi-dimensional data into semantic graph nodes, wherein the semantic graph nodes include scene nodes, reference target nodes, parking space attribute nodes, user preference nodes, and parking space nodes; The association weights between nodes in the semantic graph are quantified, and node connections between the semantic graph nodes are established based on the association weights to obtain a preset semantic graph.

4. The parking space recommendation method as described in claim 3, characterized in that, The association weights include scene association weights, and the step of quantifying the association weights between nodes in the semantic graph includes: Calculate the straight-line distance from each parking space corresponding to the parking space node to the key facility corresponding to the scene node; The number of times each parking space is entered in the scene corresponding to each scene node is counted; Based on the straight-line distance and the number of parking entries, the scene association weight between the parking space node and the scene node in the semantic graph is quantified.

5. The parking space recommendation method as described in claim 3, characterized in that, The association weights include target association weights, and the step of quantifying the association weights between nodes in the semantic graph further includes: Calculate the straight-line distance from each parking space corresponding to the parking space node to each reference target of the reference target node; Based on the straight-line distance, the target association weights between the parking space node and the scene node in the semantic graph are quantified for different semantic types, wherein the semantic types include tendency type and avoidance type.

6. The parking space recommendation method as described in claim 3, characterized in that, The association weights include attribute association weights, and the step of quantifying the association weights between nodes in the semantic graph further includes: Based on the multi-dimensional data, the matching degree between each parking space corresponding to the parking space node and the parking space attribute corresponding to the parking space attribute node is quantified. Based on the matching degree, the attribute association weight between the parking space node and the parking space attribute node in the semantic graph is quantified.

7. The parking space recommendation method as described in claim 3, characterized in that, The association weights include preference association weights, and the step of quantifying the association weights between nodes in the semantic graph further includes: Based on the number of positive and negative feedbacks from the user for each parking space corresponding to the parking space node, a specific preference matching score is calculated. Based on the preference weights of the general preference data in the multi-dimensional data, the general preference matching score of each parking space that satisfies the general preference is calculated; Based on the specific preference matching score and the general preference matching score, the preference association weight between the parking space node and the user preference node in the semantic graph is quantified.

8. The parking space recommendation method as described in claim 1, characterized in that, The step of selecting target parking space nodes corresponding to the recommendation dimension from parking space nodes in a preset semantic graph, and determining the parking space recommendation result based on the target parking space nodes, includes: Based on the recommendation dimension, the parking space nodes in the preset semantic graph are filtered and the association weight is corrected to obtain candidate parking space nodes. Target parking space nodes with a parking difficulty coefficient less than a first threshold are selected from the candidate parking space nodes; The target parking space nodes are sorted and output based on the corrected association weights to obtain the parking space recommendation results.

9. The parking space recommendation method as described in claim 8, characterized in that, The basic recommendation dimension includes automatic scene matching recommendation. The step of filtering parking space nodes in the preset semantic graph based on the recommendation dimension and correcting the association weights to obtain candidate parking space nodes includes: The scene nodes corresponding to the scene are automatically matched and recommended from the nodes of the preset semantic graph; Traverse the first association edge between the scene node and the parking space node, and filter out candidate parking space nodes whose scene association weight of the first association edge is greater than or equal to the second threshold. Based on the preference association weights corresponding to the scene preference nodes in the user preference nodes, the scene association weights of the candidate parking space nodes are corrected to obtain the corrected candidate parking space nodes.

10. The parking space recommendation method as described in claim 8, characterized in that, The basic recommendation dimension includes reference target-oriented recommendation. The step of filtering parking space nodes in the preset semantic graph based on the recommendation dimension and correcting the association weights to obtain candidate parking space nodes further includes: The reference target node corresponding to the reference target-oriented recommendation is determined from the nodes of the preset semantic graph; Traverse the second association edge between the reference target node and the parking space node to filter out candidate parking space nodes that meet the distance requirements; Based on the preference association weight of the location avoidance preference node in the user preference node, the target association weight of the candidate parking space node is corrected to obtain the corrected candidate parking space node.

11. The parking space recommendation method as described in claim 8, characterized in that, The basic recommendation dimension includes parking space attribute adaptation recommendation. The step of filtering and correcting the association weights of parking space nodes in the preset semantic graph based on the recommendation dimension to obtain candidate parking space nodes further includes: The parking space attribute nodes corresponding to the parking space attribute adaptation recommendation are determined from the nodes of the preset semantic graph; Traverse the third association edge between the parking space attribute node and the parking space node, and filter out candidate parking space nodes whose attribute association weight of the third association edge is greater than the third threshold. Based on the preference association weights of the attribute demand preference nodes in the user preference nodes, the attribute association weights of the candidate parking space nodes are corrected to obtain the corrected candidate parking space nodes.

12. The parking space recommendation method as described in claim 8, characterized in that, The step of filtering parking space nodes in the preset semantic graph based on the recommendation dimension and correcting the association weights to obtain candidate parking space nodes further includes: From the nodes of the preset semantic graph, determine the single-dimensional nodes corresponding to the single recommendation dimension in the combined recommendation dimensions; The fourth association edge between the single-dimensional node and the parking space node is traversed respectively. Based on the semantic type in the parking requirement, candidate parking space nodes that meet the parking requirement are selected. Based on the association weights of each node in the user preference node, the association weights of candidate parking space nodes in different dimensions are corrected, and the corrected association weights are summed in multiple dimensions to obtain the candidate parking space nodes after weighted summation.

13. The parking space recommendation method as described in claim 1, characterized in that, Prior to the step of receiving the parking instruction, the following is included: When a user's voice information is received, the voice information is parsed into semantic demand information to obtain demand information, and a parking instruction is generated based on the demand information. The demand information includes at least scene information, attribute information and target information. or, When the user does not issue a parking instruction, the system automatically determines the demand information, including the scene information, based on the navigation information, and generates the parking instruction based on the demand information.

14. A parking space recommendation device, characterized in that, The device includes: The response module is used to receive parking instructions and determine parking needs based on the parking instructions; The matching module is used to perform multi-dimensional data matching based on the parking demand to obtain parking space recommendation results, wherein the multi-dimensional data includes multiple items from static map data, user preference data, and real-time dynamic data; The response module includes: The calculation submodule is used to calculate the number of times the data attributes of the behavioral data during the vehicle parking process overlap with the data attributes of the general preference data, wherein the behavioral data includes at least one of active selection data, active feedback data, and system inference data; based on the data type of the general preference data, the overlap number is classified to obtain the target overlap number corresponding to different data types; based on the target overlap number and the differential weights corresponding to different data types, the preference score of the general preference data of different data types is calculated; based on the preference score, the updated dynamic weight of the general preference data of different data types is calculated, wherein the dynamic weight is used to update the preference weight of the general preference data, and the preference weight of the general preference data is used to calculate the preference association weight between parking space nodes and user preference nodes in a preset semantic graph; The matching module includes: The filtering submodule is used to determine the recommendation dimension corresponding to the parking demand, wherein the recommendation dimension includes any one of the basic recommendation dimension and the combined recommendation dimension; from the parking space nodes of the preset semantic graph, the target parking space node corresponding to the recommendation dimension is obtained, and the parking space recommendation result is determined based on the target parking space node.

15. A parking space recommendation device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the parking space recommendation method as described in any one of claims 1 to 13.

16. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the parking space recommendation method as described in any one of claims 1 to 13.

17. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the parking space recommendation method as described in any one of claims 1 to 13.

Images