Parking lot recommendation method and recommendation device, target vehicle, and storage medium

Abstract

The present invention discloses a parking lot recommendation method and recommendation device, a target vehicle, and a storage medium. The method includes: obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle, wherein the location information is used to search for the parking lot information to be analyzed; combining the parking lot information, the user information, and the recommended parking relationship into a site recommendation triple; inputting the site recommendation triple into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the site recommendation triple into an item vector and a relationship vector; substituting the item vector and the relationship vector into a preset recommendation algorithm to determine the parking lot vector and the user vector; and outputting the parking lot recommendation information based on the parking lot vector and the user vector. The present invention solves the technical problem in the related art that errors are prone to occur when recommending vehicle parking lots, resulting in a lack of recommendation interaction information and low accuracy.

Description

Technical Field

[0001] The present invention relates to the field of information processing technology, and in particular to a parking lot recommendation method and device, a target vehicle, and a storage medium. Background Art

[0002] In the related art, with the increasing popularity of family cars and commercial vehicles, a large number of vehicles are faced with the problem of how to find a suitable parking space. Various map software or parking lot recommendation software use different recommendation algorithms to recommend parking spaces.

[0003] The current traditional recommendation algorithm adopts collaborative filtering recommendation algorithm, which generally adopts nearest neighbor technology. It first uses the user's historical preferences to calculate the distance between users, and then finds the nearest neighbor user. It judges the user's preference for the product based on the neighbor user's evaluation and weighted value of the preference. The backend system recommends to users based on the degree of preference. The premise of recommending to users is to find other users with similar interests as the user, and make recommendations based on the interests of other users. This recommendation can sort out simple relationships and can easily handle unstructured objects that do not have complex information. However, in the face of increasingly complex content, increasing users and more complex nodes, it is necessary to solve its sparsity and cold start problems. For example, a user may have been to only a few parking lots. Using such a small amount of data to infer a large amount of unknown information will cause errors in the algorithm, resulting in a lack of interactive information for recommendations and low accuracy.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present invention provide a parking lot recommendation method and recommendation device, a target vehicle, and a storage medium, so as to at least solve the technical problem in the related art that errors are prone to occur when recommending parking lots for vehicles, resulting in a lack of recommendation interactive information and low accuracy.

[0006] According to one aspect of an embodiment of the present invention, a method for recommending a parking space is provided, comprising: obtaining location information of a current vehicle and user information of a target user driving the current vehicle, wherein the location information is used to search for parking space information to be analyzed; combining the parking space information, the user information and the recommended parking relationship into a venue recommendation triplet; inputting the venue recommendation triplet into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the venue recommendation triplet into an item vector and a relationship vector; substituting the item vector and the relationship vector into a preset recommendation algorithm to determine a parking space vector and a user vector; and outputting parking space recommendation information based on the parking space vector and the user vector.

[0007] Optionally, the step of obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle includes: responding to a parking instruction, obtaining the positioning signal of the current vehicle to determine the location information of the current vehicle; obtaining the user occupation information, hobby information, and age information of the target user to determine the user information, wherein the user information is used to evaluate the consumption level of the target user and the price range accepted by the parking lot.

[0008] Optionally, after obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle, it also includes: obtaining the social network information and relative information of the target user; based on the social network information and the relative information, outputting the recommended weight parameters of the parking lot, wherein the recommended weight parameters affect the parking lot recommendation information to be output.

[0009] Optionally, before inputting the venue recommendation triplet into a pre-trained knowledge graph model, it also includes: obtaining an initial knowledge graph network constructed within a historical preset time period; supplementing the information graph of parking lot information and user information extracted during the current time period to the initial knowledge graph network to build a target knowledge graph; and constructing the knowledge graph model in combination with the target knowledge graph.

[0010] Optionally, the step of outputting parking recommendation information based on the parking space vector and the user vector includes: obtaining a first vector label corresponding to the parking space vector and a second vector label corresponding to the user vector; calculating a vector distance parameter based on the parking space vector and the corresponding first vector label, the user vector and the corresponding second vector label; and outputting the parking space recommendation information in combination with the vector distance parameter.

[0011] Optionally, the step of outputting parking recommendation information in combination with the vector distance parameter includes: analyzing the target user's preference parameters for each recommended parking lot in the parking lot information to be analyzed based on the vector distance parameter; and outputting the parking recommendation information in combination with the preference parameters.

[0012] According to another aspect of an embodiment of the present invention, a parking lot recommendation device is also provided, including: an acquisition unit, used to acquire location information of a current vehicle and user information of a target user driving the current vehicle, wherein the location information is used to search for parking lot information to be analyzed; a combination unit, used to combine the parking lot information, the user information and the recommended parking relationship into a venue recommendation triplet; an input unit, used to input the venue recommendation triplet into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the venue recommendation triplet into an item vector and a relationship vector; a substitution unit, used to substitute the item vector and the relationship vector into a preset recommendation algorithm to determine a parking lot vector and a user vector; and an output unit, used to output parking lot recommendation information based on the parking lot vector and the user vector.

[0013] Optionally, the acquisition unit includes: a first response module, used to respond to a parking instruction and acquire the positioning signal of the current vehicle to determine the location information of the current vehicle; a first acquisition module, used to acquire the user occupation information, hobby information, and age information of the target user to determine the user information, wherein the user information is used to evaluate the consumption level of the target user and the price range accepted by the parking lot.

[0014] Optionally, the parking lot recommendation device also includes: a second acquisition module, used to obtain social network information and relative information of the target user after obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle; a first output module, used to output a recommendation weight parameter of the parking lot based on the social network information and the relative information, wherein the recommendation weight parameter affects the parking lot recommendation information to be output.

[0015] Optionally, the parking lot recommendation device also includes: a third acquisition module, used to obtain an initial knowledge graph network constructed within a historical preset time period before inputting the venue recommendation triplet into a pre-trained knowledge graph model; a supplement module, used to supplement the information graph of parking lot information and user information extracted in the current time period to the initial knowledge graph network to build a target knowledge graph; and a construction module, used to construct the knowledge graph model in combination with the target knowledge graph.

[0016] Optionally, the output unit includes: a fourth acquisition module, used to acquire a first vector label corresponding to the parking space vector and a second vector label corresponding to the user vector; a calculation module, used to calculate a vector distance parameter based on the parking space vector and the corresponding first vector label, the user vector and the corresponding second vector label; and a second output module, used to output parking space recommendation information in combination with the vector distance parameter.

[0017] Optionally, the second output module includes: an analysis submodule, used to analyze the target user's preference parameters for each recommended parking lot in the parking lot information to be analyzed based on the vector distance parameters; and an output submodule, used to output parking lot recommendation information in combination with the preference parameters.

[0018] According to another aspect of an embodiment of the present invention, a target vehicle is provided, comprising: a processor; and a memory for storing a computer program of the processor; wherein the processor is configured to execute any one of the above-mentioned methods for recommending a parking space by executing the computer program.

[0019] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is further provided, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute any of the above-mentioned methods for recommending parking spaces.

[0020] The recommendation algorithm of the present invention uses a knowledge graph model to form a large knowledge graph network of user basic information and parking lot information, which can be understood as binding and extending existing information content to make the interactive information between users and parking lots become close.

[0021] The recommendation algorithm in the present invention connects users and users and users and parking lots to form a large social network, which enables interaction between nodes and makes the recommendation explainable.

[0022] In the present invention, the location information of the current vehicle and the user information of the target user driving the current vehicle are obtained, wherein the location information is used to search for the parking lot information to be analyzed, the parking lot information, the user information and the recommended parking relationship are combined into a site recommendation triple, and the site recommendation triple is input into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the site recommendation triple into an item vector and a relationship vector, and the item vector and the relationship vector are substituted into a preset recommendation algorithm to determine the parking lot vector and the user vector, and the parking lot recommendation information is output based on the parking lot vector and the user vector. In the present application, the parking lot information, the user information and the recommended parking relationship are combined into a site recommendation triple, and then the user basic information and the parking lot information are formed into a large knowledge graph network through the knowledge graph model, and the existing information content is bound and extended, so that the interactive information between the user and the parking lot becomes close, and the recommendation accuracy is improved, thereby solving the technical problem that errors are prone to occur when recommending vehicle parking lots in the related technology, resulting in a lack of recommended interactive information and low accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0024] Figure 1 is a flow chart of an optional parking lot recommendation method according to an embodiment of the present invention;

[0025] Figure 2 is a schematic diagram of extracting user social network information according to an embodiment of the present invention;

[0026] Figure 3 is a schematic diagram of an optional parking lot recommendation method based on knowledge graph according to an embodiment of the present invention;

[0027] Figure 4 is a schematic diagram of an optional parking lot recommendation device according to an embodiment of the present invention;

[0028] Figure 5 The present invention is a hardware structure block diagram of an electronic device (or mobile device) for recommending a parking space according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] The present invention can be applied to various scenarios such as navigation systems / software, map software, vehicle control systems, parking lot management systems, etc., and involves the field of parking lot recommendation. The purpose of the present invention is to utilize the user's basic information and parking records, combined with information such as the location, price, and remaining parking spaces of nearby parking lots, to realize a parking lot recommendation algorithm based on a knowledge graph neural network, and to make interpretable and accurate recommendations based on the user's basic information and parking lot information.

[0032] The parking lot recommendation algorithm in the present invention can build user and parking lot information into a ternary graph in the form of node-association-node. Large-scale data forms small ternary graphs, and the small ternary graphs are formed into a large knowledge graph network through the association between two nodes. The knowledge graph has natural high dimensionality and heterogeneity. By processing it through knowledge graph feature learning, low-dimensional vector representations between entities and relationships can be obtained. These low-dimensional vectors are combined with the recommendation algorithm to give corresponding recommendation results.

[0033] The present invention is described in detail below in conjunction with various embodiments.

[0034] Embodiment 1

[0035] According to an embodiment of the present invention, a method embodiment for recommending a parking space is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer programs, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0036] Figure 1 is a flow chart of an optional parking lot recommendation method according to an embodiment of the present invention, such as Figure 1 As shown, the method comprises the following steps:

[0037] Step S101, obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle, wherein the location information is used to search for parking space information to be analyzed;

[0038] Step S102, combining parking space information, user information and recommended parking relationship into a site recommendation triplet;

[0039] Step S103, inputting the venue recommendation triplet into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the venue recommendation triplet into an item vector and a relationship vector;

[0040] Step S104, substituting the item vector and the relationship vector into a preset recommendation algorithm to determine a parking space vector and a user vector;

[0041] Step S105: output parking lot recommendation information based on the parking lot vector and the user vector.

[0042] Through the above steps, the location information of the current vehicle and the user information of the target user driving the current vehicle can be obtained, wherein the location information is used to search for the parking lot information to be analyzed, the parking lot information, the user information and the recommended parking relationship are combined into a venue recommendation triple, and the venue recommendation triple is input into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the venue recommendation triple into an item vector and a relationship vector, substitute the item vector and the relationship vector into a preset recommendation algorithm to determine the parking lot vector and the user vector, and output the parking lot recommendation information based on the parking lot vector and the user vector. In this embodiment, the parking lot information, the user information and the recommended parking relationship are combined into a venue recommendation triple, and then the user basic information and the parking lot information are formed into a large knowledge graph network through the knowledge graph model, and the existing information content is bound and extended, so that the interactive information between the user and the parking lot becomes close, and the recommendation accuracy is improved, thereby solving the technical problem that errors are prone to occur when recommending vehicle parking lots in related technologies, resulting in a lack of recommended interactive information and low accuracy.

[0043] The present invention is described in detail below in conjunction with the above-mentioned implementation steps.

[0044] Step S101, obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle, wherein the location information is used to search for parking space information to be analyzed.

[0045] The type of the current vehicle includes but is not limited to: new energy vehicles (including but not limited to: electric vehicles, hybrid vehicles), commercial vehicles, school buses, public buses, and sports cars; a displayable or playable screen is provided in the driving area or control area of ​​the current vehicle, and the parking lot recommendation information can be presented through the screen. At the same time, the current vehicle can also be configured with a loudspeaker, a player, etc., and after the user selects a recommended parking lot, the voice of the navigation planning recommendation is played to assist the user in driving the current vehicle to the recommended parking lot.

[0046] When recommending parking spaces for the current vehicle, it is necessary not only to consider the location information and the information of the user driving the current vehicle, but also to consider the vehicle's own information, such as vehicle height, vehicle width, vehicle length, and the number of vehicle seats. At the same time, it is necessary to consider the planning information of the parking space, for example, whether it is possible to park buses, coaches, and tourist coaches that are too long.

[0047] Optionally, the step of obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle includes: responding to a parking instruction, obtaining the positioning signal of the current vehicle to determine the location information of the current vehicle; obtaining the user occupation information, hobby information, and age information of the target user to determine the user information, wherein the user information is used to evaluate the consumption level of the target user and the acceptable price range of the parking lot.

[0048] In this embodiment, the parking instruction can be issued by the driver or passenger of the current vehicle. The types of the parking instruction include but are not limited to: touch instructions (for example, the user's finger touches the parking recommendation button box), voice commands (for example, the user issues "find a parking lot"), and autonomous recommendation instructions (that is, when the software finds that it is about to reach the destination according to the navigation information, it issues the parking instruction to find a suitable parking lot).

[0049] In an embodiment of the present invention, the user's portrait information can be collected through videos, images, etc. After obtaining the parking instruction / signal of the user's desire to park, the positioning information of the user's location is fed back, and the user's portrait information is obtained from the database, including the user's age, occupation, interests and hobbies, salary, occupation, etc., and then the parking fee and location information of parking lots within a certain range (for example, within two kilometers nearby) are obtained, as well as information about parking lots that the user has visited in the past.

[0050] After obtaining the positioning information returned by the system, the information of parking lots within a certain range (for example, within two kilometers) is obtained from the database. If there is no parking lot, the specified distance is increased (for example, one kilometer is increased) until a parking lot is found, and then it is determined whether there are vacancies in the parking lot. If there are no vacancies, the search range is further increased until a parking lot is found with vacancies, and the parking lot information is returned to the system.

[0051] It should be noted that after obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle, it also includes: obtaining the social network information and relative information of the target user; based on the social network information and relative information, outputting the recommended weight parameters of the parking lot, wherein the recommended weight parameters affect the parking lot recommendation information to be output.

[0052] The user's social network information is a core information source for recommendation algorithms. For newly added users or items, the system may not be able to accurately model and recommend them because it has no historical interaction information. This requires the introduction of additional auxiliary information as input to help modeling. If a user often goes to a nearby parking lot, his friends may also prefer this parking lot. Therefore, the user's social network information and relative information can be obtained to output the recommendation weight parameters of the parking lot. The user may also choose to go to the parking lot that his friends or relatives often go to.

[0053] Figure 2 is a schematic diagram of extracting user social network information according to an embodiment of the present invention, such as Figure 2 As shown, the old user and the new user are friends. If the white-collar worker and the old user both refer to the parking lot price ( Figure 2 5 yuan / hour), then it will also likely choose a parking lot with that price. The information of old users includes occupation ( Figure 2 White-collar workers are used as an example) and age ( Figure 2 (The information is provided by the Chinese Youth Association.)

[0054] Figure 2 As shown in the figure, when a new user transmits parking information to the system, if there is a social information circle, the new user information will be put into the established knowledge graph, and the new user and old users will have information interaction. The solid line represents the existing relationship, and the dotted line represents the possible relationship, which effectively solves the problem of scarce information interaction between new users and the system.

[0055] Step S102: combining parking space information, user information and recommended parking relationships into a site recommendation triplet.

[0056] The knowledge graph is currently a very effective deep learning algorithm model. Its ontology is a large semantic network. Each node represents an entity, and the edge represents the semantic relationship between entities. It can include small triples, which can be in the form of entity-relationship-entity. For example, parking in parking lot A costs three yuan per hour, where parking lot A and three yuan / hour are two entities, and parking fee is a relationship.

[0057] The knowledge graph provides rich semantics for the parking lot recommendation algorithm. Applying the knowledge graph to the parking lot recommendation algorithm can provide the algorithm with authenticity, explainability, and depth.

[0058] Optionally, before inputting the venue recommendation triplet into the pre-trained knowledge graph model, it also includes: obtaining an initial knowledge graph network constructed within a historical preset time period; supplementing the information graph of parking lot information and user information extracted during the current time period to the initial knowledge graph network to build a target knowledge graph; and constructing a knowledge graph model in combination with the target knowledge graph.

[0059] In this embodiment, the knowledge graph and the parking recommendation algorithm can be combined. In the process of combination, the feature learning of the knowledge graph is used. The knowledge graph feature learning learns each entity and relationship into a low-dimensional vector space without destroying the original relationship. Feature learning can use Euclidean space, mainly mapping the relationship to a vector or matrix space or capturing the interaction of the relationship.

[0060] In this embodiment, the initial knowledge graph network can learn the information graph of parking lot information and user information extracted in the current time period, and then supplement these information graphs to the initial knowledge graph network to build the target knowledge graph. The initial knowledge graph can also be understood as a historical knowledge graph. The knowledge graph constructed using historical information has not been updated in real time. In this embodiment, the information graph of parking lot information and user information extracted in the current time period can be supplemented to the initial knowledge graph network, and the knowledge graph can be updated in real time to obtain the target knowledge graph.

[0061] Step S103, inputting the venue recommendation triplet into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the venue recommendation triplet into an item vector and a relationship vector.

[0062] It should be noted that this embodiment can use knowledge graph feature learning to map a triple into a low-dimensional vector. The triple cannot be used for calculation, but the vector can be used as a computable symbol. It is used as an important input of our algorithm, which reduces the heterogeneity and complexity of the knowledge graph and the computational burden brought by the knowledge graph. We use the triple as input to obtain the parking knowledge graph, perform knowledge graph feature learning, obtain item vectors and relationship vectors, and then introduce these low-dimensional vectors into the recommendation algorithm to obtain the parking lot vector and user vector we need.

[0063] Step S104: Substitute the item vector and the relationship vector into a preset recommendation algorithm to determine a parking space vector and a user vector.

[0064] Step S105: output parking lot recommendation information based on the parking lot vector and the user vector.

[0065] Optionally, the step of outputting parking recommendation information based on the parking space vector and the user vector includes: obtaining a first vector label corresponding to the parking space vector and a second vector label corresponding to the user vector; calculating a vector distance parameter based on the parking space vector and the corresponding first vector label, the user vector and the corresponding second vector label; and outputting the parking space recommendation information in combination with the vector distance parameter.

[0066] In this example, the model labels and weights each vector for subsequent distance calculation. The specific labels are shown in Table 1 below:

[0067] Table 1 Labeling and weighting vectors

[0068]

[0069] The larger the label value, the greater the influence on the user's preference. In this embodiment, labeling of the vector can be defined according to the number of entity relationships of the vector. The specific label value formula is as follows:

[0070]

[0071] Among them, a: the total value of the defined label x: the correlation coefficient.

[0072] Then, using the traditional Euclidean distance to calculate the vector distance will result in errors with the actual situation. This embodiment improves it by introducing labels to optimize the algorithm of this embodiment:

[0073] Euclidean distance algorithm:

[0074] in, Tag value.

[0075] It should be noted that the step of outputting parking recommendation information in combination with the vector distance parameter includes: analyzing the target user's preference parameters for each recommended parking lot in the parking lot information to be analyzed based on the vector distance parameter; and outputting the parking recommendation information in combination with the preference parameters.

[0076] Based on the obtained low-dimensional vector, the user's preference for each parking lot (or preference parameter\preference value) can be calculated through the formula. The larger the value, the closer the distance between the two, and the more likely the user is to prefer the parking lot. The model is built based on the knowledge graph based on the parking lot information and user portrait information, and then the low-dimensional vector space synthesized by the user portrait information and the parking lot information is formed. This model forms a large knowledge graph network between the user and the parking lot. There is a definite relationship between the user and the parking lot, making the recommendation results explainable.

[0077] Through the above embodiments, the user basic information and parking lot information can be formed into a large knowledge graph network through the knowledge graph model, so that the interactive information between users and parking lots becomes close. At the same time, in this embodiment, users and users as well as users and parking lots are connected to form a large social network, which enables interaction between nodes and makes recommendations explainable.

[0078] The present invention is described in detail below in conjunction with another optional embodiment.

[0079] Embodiment 2

[0080] In this embodiment, a parking lot recommendation algorithm is taken as an example for illustration.

[0081] Figure 3It is a schematic diagram of an optional parking lot recommendation method based on knowledge graph according to an embodiment of the present invention. In order to realize the key module information collection, knowledge graph construction, recommendation algorithm processing and output results of the parking lot recommendation algorithm based on knowledge graph, the implementation process includes: s1 / s2-m1-m2-m3-m4-m5-m6-m7-m8;

[0082] like Figure 3 As shown, the algorithm implementation process is as follows:

[0083] Process s1 / s2: user portrait information collection (including basic information such as user gender, age, occupation, hobbies, salary, etc.) and location information collection; parking lot information collection (current parking fees and distance information of nearby parking lots, historical parking information (including historical parking lot usage fees, etc.)).

[0084] After receiving a signal that the user wants to park, the system will feedback the user's location information (for example, GPS location information or Beidou location information), obtain the user's portrait information from the database, including the user's age, occupation, interests and hobbies, salary, occupation, etc., and then obtain the parking fee and location information of parking lots within a certain range nearby (such as within two kilometers), as well as information about parking lots the user has visited in the past.

[0085] Process m1: Basic information data. After processing the user portrait information, basic information data can be obtained.

[0086] The current location is returned through the on-board signal of the current vehicle to search for suitable parking lots nearby. The user's portrait information is obtained through mobile app registration or other means, and the required information is extracted and stored, including age, occupation, interests and hobbies, salary, and occupation, so as to obtain the user's consumption level and acceptable parking price range.

[0087] The training data of the model is based on the user portrait information and the m1 data processed by the current location.

[0088] Process m2: Collection of user social circle information data.

[0089] User social networks are a core source of information for recommendation algorithms. For newly added users or items, it may be impossible to accurately model and recommend them because the system has no historical interaction information. This requires the introduction of additional auxiliary information as input to help modeling. For example, if a user often goes to a nearby parking lot, his friends may also prefer this parking lot.

[0090] like Figure 2As shown in the figure, when a new user transmits parking information to the system, if there is a social information circle, the new user information will be put into the established knowledge graph, and the new user and old users will have information interaction. The solid line represents the existing relationship, and the dotted line represents the possible relationship, which effectively solves the problem of scarce information interaction between new users and the system.

[0091] Process m3: parking lot usage information.

[0092] After obtaining the positioning location returned by the system, the information of parking lots within a certain range (for example, two kilometers) nearby can be obtained from the database. If there is no parking lot, the specified distance value (for example, one kilometer) is continuously increased until a parking lot is found, and a determination is made as to whether there are vacancies in the parking lot. If there are no vacancies, the search range is continuously increased until a parking lot is found with vacancies, and the parking lot information is then returned to the system.

[0093] Process m4: Historical knowledge graph acquisition (history).

[0094] The construction of knowledge graph is the core content of this recommendation algorithm. The following is an explanation of the model principle:

[0095] The knowledge graph is currently a very effective deep learning algorithm model. Its ontology is a large semantic network. Each node represents an entity, and the edge represents the semantic relationship between entities. It is composed of small (entity, relationship, entity) triplets.

[0096] For example, the triple (Parking Lot A - Parking Fee - 3 Yuan / Hour) indicates that the parking fee at Parking Lot A is 3 Yuan per hour, where Parking Lot A and 3 Yuan / Hour are two entities and Parking Fee is a relationship.

[0097] The knowledge graph provides rich semantics for the parking lot recommendation algorithm. Applying the knowledge graph to the parking lot recommendation algorithm can provide the algorithm with authenticity, explainability, and depth.

[0098] Process m5: Information integration.

[0099] The acquired basic information is converted into the required data format through NLP natural language processing and input into the algorithm model as a signal.

[0100] Process m6: Model deployment.

[0101] This module mainly associates the original knowledge graph network with the newly added small knowledge graph to build a larger knowledge graph, provide model algorithms for the next module, and output vector features for the next module.

[0102] Process m7: Parking lot knowledge graph (real time).

[0103] In this embodiment, in order to combine the knowledge graph and the parking recommendation algorithm, feature learning of the knowledge graph is used. The knowledge graph feature learning learns each entity and relationship into a low-dimensional vector space without destroying the original relationship. Feature learning can use Euclidean space, which mainly maps the relationship to vector or matrix space or captures the interaction of relationships.

[0104] It should be noted that there are many algorithms and strategies for feature learning. In this embodiment, the TransE algorithm is mainly used to implement knowledge graph feature learning. Take a triple example (Hongqi HS5, car model, SUV). The TransE algorithm is to map this triple into a low-dimensional vector, so that Hongqi HS5 is not just a name, but becomes a low-dimensional vector, such as [0.2, 1.5, 0.4, 1, 8]. The algorithm believes that a triple can represent the head entity + relationship = tail entity. A correct triple head entity plus a relationship can be represented as a tail entity, and an incorrect triple cannot be represented as such a vector. In this embodiment, a function d(x, y) can be defined to represent the relationship between the head and tail entities, which is expressed in this algorithm as: d(head entity, relationship, tail entity) = || head entity + relationship - tail entity

[0105] In this embodiment, knowledge graph feature learning can be used to map a triple into a low-dimensional vector. The triple cannot be used for calculation, but the vector can be used as a computable symbol. It is used as an important input of the algorithm, which reduces the heterogeneity and complexity of the knowledge graph and the computational burden brought by the knowledge graph. We use the triple as input to obtain the parking knowledge graph, perform knowledge graph feature learning, obtain item vectors and relationship vectors, and then introduce these low-dimensional vectors into the recommendation algorithm to obtain the required parking vectors and user vectors.

[0106] Process m8: Explainable parking lot recommendation results.

[0107] Based on the obtained low-dimensional vector, the model can calculate the user's preference for each parking lot through a formula. The larger the value, the closer the distance between the two, and the more likely the user is to prefer the parking lot. The model is built based on the knowledge graph based on parking lot information and user portrait information. It is a low-dimensional vector space that synthesizes user portrait information and parking lot information. This model forms a large knowledge graph network between users and parking lots. There is a definite relationship between users and parking lots, making the recommendation results explainable.

[0108] In this embodiment, the user and parking lot information can be constructed into a ternary graph in the form of node-association-node. Large-scale data forms small ternary graphs, and the small ternary graphs are formed into a large knowledge graph network through the association between two nodes. The knowledge graph has natural high dimensionality and heterogeneity. By processing it through knowledge graph feature learning, low-dimensional vector representations between entities and relationships can be obtained. These low-dimensional vectors are combined with recommendation algorithms to give corresponding recommendation results.

[0109] In this embodiment, users and users, users and parking lots are connected to form a large social network knowledge graph, which enables interaction between nodes and makes the recommendations explainable.

[0110] The present invention is described below in conjunction with another optional embodiment.

[0111] Embodiment 3

[0112] This embodiment provides a parking lot recommendation device, and each implementation unit included in the recommendation device corresponds to each implementation step in the above-mentioned embodiment 1.

[0113] Figure 4 is a schematic diagram of an optional parking lot recommendation device according to an embodiment of the present invention, such as Figure 4 As shown, the recommendation device may include: an acquisition unit 41, a combination unit 43, an input unit 45, a substitution unit 47, and an output unit 49, wherein:

[0114] An acquisition unit 41 is used to acquire the location information of the current vehicle and the user information of the target user driving the current vehicle, wherein the location information is used to search for parking lot information to be analyzed;

[0115] A combining unit 43, used to combine the parking lot information, the user information and the recommended parking relationship into a parking lot recommendation triplet;

[0116] An input unit 45, used to input the venue recommendation triplet into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the venue recommendation triplet into an item vector and a relationship vector;

[0117] A substitution unit 47, used to substitute the item vector and the relationship vector into a preset recommendation algorithm to determine a parking space vector and a user vector;

[0118] The output unit 49 is used to output parking lot recommendation information based on the parking lot vector and the user vector.

[0119] The above-mentioned parking lot recommendation device can obtain the location information of the current vehicle and the user information of the target user driving the current vehicle through the acquisition unit 41, wherein the location information is used to search for the parking lot information to be analyzed, and the parking lot information, user information and recommended parking relationship are combined into a venue recommendation triple through the combination unit 43, and the venue recommendation triple is input into a pre-trained knowledge graph model through the input unit 45, wherein the knowledge graph model is used to map the venue recommendation triple into an item vector and a relationship vector, and the item vector and the relationship vector are substituted into a preset recommendation algorithm through the substitution unit 47 to determine the parking lot vector and the user vector, and the parking lot recommendation information is output based on the parking lot vector and the user vector through the output unit 49. In this embodiment, parking lot information, user information and recommended parking relationships are combined into a venue recommendation triple, and then the user basic information and parking lot information are formed into a large knowledge graph network through the knowledge graph model, and the existing information content is bound and extended, so that the interactive information between the user and the parking lot becomes close, and the recommendation accuracy is improved, thereby solving the technical problem that errors are prone to occur when recommending vehicle parking lots in related technologies, resulting in a lack of recommended interactive information and low accuracy.

[0120] Optionally, the acquisition unit includes: a first response module, used to respond to the parking instruction and acquire the positioning signal of the current vehicle to determine the location information of the current vehicle; a first acquisition module, used to acquire the user occupation information, hobby information, and age information of the target user to determine the user information, wherein the user information is used to evaluate the consumption level of the target user and the acceptable price range of the parking lot.

[0121] Optionally, the parking lot recommendation device also includes: a second acquisition module, used to obtain social network information and relative information of the target user after obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle; a first output module, used to output a recommendation weight parameter of the parking lot based on the social network information and the relative information, wherein the recommendation weight parameter affects the parking lot recommendation information to be output.

[0122] Optionally, the parking lot recommendation device also includes: a third acquisition module, used to obtain an initial knowledge graph network constructed within a historical preset time period before inputting the venue recommendation triplet into a pre-trained knowledge graph model; a supplement module, used to supplement the information graph of parking lot information and user information extracted in the current time period to the initial knowledge graph network to build a target knowledge graph; and a construction module, used to construct a knowledge graph model in combination with the target knowledge graph.

[0123] Optionally, the output unit includes: a fourth acquisition module, used to obtain a first vector label corresponding to the parking space vector and a second vector label corresponding to the user vector; a calculation module, used to calculate a vector distance parameter based on the parking space vector and the corresponding first vector label, the user vector and the corresponding second vector label; and a second output module, used to output parking space recommendation information in combination with the vector distance parameter.

[0124] Optionally, the second output module includes: an analysis submodule for analyzing the target user's preference parameters for each recommended parking lot in the parking lot information to be analyzed based on the vector distance parameters; and an output submodule for outputting the parking lot recommendation information in combination with the preference parameters.

[0125] According to another aspect of an embodiment of the present invention, a target vehicle is provided, comprising: a processor; and a memory for storing a computer program of the processor; wherein the processor is configured to execute any of the above-mentioned methods for recommending a parking space by executing the computer program.

[0126] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium including a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute any of the above-mentioned methods for recommending parking spaces.

[0127] The above-mentioned parking lot recommendation device may also include a processor and a memory. The above-mentioned acquisition unit 41, combination unit 43, input unit 45, substitution unit 47, output unit 49, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0128] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the item vector and the relationship vector are substituted into the preset recommendation algorithm by adjusting the kernel parameters to determine the parking space vector and the user vector, and output the parking space recommendation information based on the parking space vector and the user vector.

[0129] The above-mentioned memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one storage chip.

[0130] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that initializes the following method steps: obtaining location information of a current vehicle and user information of a target user driving the current vehicle, wherein the location information is used to search for parking space information to be analyzed; combining the parking space information, user information, and recommended parking relationships into a venue recommendation triplet; inputting the venue recommendation triplet into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the venue recommendation triplet into an item vector and a relationship vector; substituting the item vector and the relationship vector into a preset recommendation algorithm to determine a parking space vector and a user vector; and outputting parking space recommendation information based on the parking space vector and the user vector.

[0131] Figure 5 FIG. 1 is a hardware structure block diagram of an electronic device (or mobile device) for recommending a parking lot according to an embodiment of the present invention. Figure 5 As shown, the electronic device may include one or more (102a, 102b, ..., 102n are used to illustrate) processors 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), and a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a keyboard, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 5 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 5 More or fewer components as shown, or with Figure 5 Different configurations are shown.

[0132] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0133] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0134] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0135] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0136] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0137] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0138] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for recommending parking spaces, characterized in that: include: Acquire location information of a current vehicle and user information of a target user driving the current vehicle, wherein the location information is used to search for parking space information to be analyzed; Combining the parking lot information, the user information and the recommended parking relationship into a site recommendation triplet; Inputting the venue recommendation triple into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the venue recommendation triple into an item vector and a relationship vector; Substituting the item vector and the relationship vector into a preset recommendation algorithm to determine a parking space vector and a user vector; Outputting parking recommendation information based on the parking space vector and the user vector, including: acquiring a first vector label corresponding to the parking space vector and a second vector label corresponding to the user vector; calculating a vector distance parameter based on the parking space vector and the corresponding first vector label, the user vector and the corresponding second vector label; and outputting the parking space recommendation information in combination with the vector distance parameter; The step of outputting parking recommendation information in combination with the vector distance parameter comprises: analyzing the target user's preference parameters for each recommended parking lot in the parking lot information to be analyzed based on the vector distance parameter; and outputting the parking recommendation information in combination with the preference parameters.

2. The recommendation method according to claim 1, characterized in that: The step of obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle includes: In response to the parking instruction, obtaining the positioning signal of the current vehicle to determine the position information of the current vehicle; The user occupation information, hobby information, and age information of the target user are obtained to determine the user information, wherein the user information is used to evaluate the consumption level of the target user and the acceptable price range of the parking lot.

3. The recommendation method according to claim 2, characterized in that: After obtaining the location information of the current vehicle and the user information of the target user driving the current vehicle, the method further includes: Obtaining social network information and relative information of the target user; Based on the social network information and the relative information, a recommendation weight parameter of a parking space is output, wherein the recommendation weight parameter affects the parking space recommendation information to be output.

4. The recommendation method according to claim 1, characterized in that: Before inputting the venue recommendation triple into the pre-trained knowledge graph model, the method further includes: Obtain the initial knowledge graph network constructed within a preset historical time period; Supplementing the information graph of parking lot information and user information extracted in the current time period to the initial knowledge graph network to build a target knowledge graph; Combined with the target knowledge graph, the knowledge graph model is constructed.

5. A parking lot recommendation device, characterized in that: include: an acquisition unit, used to acquire location information of a current vehicle and user information of a target user driving the current vehicle, wherein the location information is used to search for parking space information to be analyzed; A combining unit, configured to combine the parking lot information, the user information and the recommended parking relationship into a site recommendation triplet; An input unit, used to input the venue recommendation triple into a pre-trained knowledge graph model, wherein the knowledge graph model is used to map the venue recommendation triple into an item vector and a relationship vector; A substitution unit, used to substitute the item vector and the relationship vector into a preset recommendation algorithm to determine a parking space vector and a user vector; an output unit, configured to output parking lot recommendation information based on the parking lot vector and the user vector, The output unit includes: a fourth acquisition module, used to acquire a first vector label corresponding to the parking space vector and a second vector label corresponding to the user vector; a calculation module, used to calculate a vector distance parameter based on the parking space vector and the corresponding first vector label, the user vector and the corresponding second vector label; a second output module, used to output parking space recommendation information in combination with the vector distance parameter; The second output module includes: an analysis submodule for analyzing the target user's preference parameters for each recommended parking lot in the parking lot information to be analyzed based on the vector distance parameter; and an output submodule for outputting parking lot recommendation information in combination with the preference parameters.

6. The recommendation device according to claim 5, characterized in that: The acquisition unit comprises: A first response module is used to respond to the parking instruction and obtain the positioning signal of the current vehicle to determine the position information of the current vehicle; The first acquisition module is used to acquire the user occupation information, hobby information, and age information of the target user to determine the user information, wherein the user information is used to evaluate the consumption level of the target user and the acceptable price range of the parking lot.

7. A target vehicle, characterized in that: include: processor; as well as a memory for storing a computer program for the processor; The processor is configured to execute the parking space recommendation method according to any one of claims 1 to 4 by executing the computer program.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein the computer program is executed by a computer to perform the parking space recommendation method according to any one of claims 1 to 4.

Images