CN112989194B - Recommendation method and system integrating user request and service preference of Internet of vehicles - Google Patents
Recommendation method and system integrating user request and service preference of Internet of vehicles Download PDFInfo
- Publication number
- CN112989194B CN112989194B CN202110290783.1A CN202110290783A CN112989194B CN 112989194 B CN112989194 B CN 112989194B CN 202110290783 A CN202110290783 A CN 202110290783A CN 112989194 B CN112989194 B CN 112989194B
- Authority
- CN
- China
- Prior art keywords
- service
- recommendation
- services
- user
- vehicle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Computational Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Algebra (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a recommendation method and a recommendation system for integrating user requests and service preferences of internet of vehicles, which provide information release service for internet-connected vehicles through two stages of online recommendation and offline training. The online recommendation stage acquires the service requirements of the online vehicles in real time, efficiently releases the service end information to corresponding online vehicle users through a collaborative filtering recommendation algorithm of user requirement and service preference clustering, and synchronously stores the service end information into a user preference record; and in the off-line stage, the service preference data of the user is acquired and fused, calculation is carried out through a user service preference model, and the issued information is synchronously updated. The invention can optimize the information transmission time delay and improve the service quality of the Internet of vehicles service recommendation system.
Description
Technical Field
The invention relates to a recommendation method and a recommendation system for integrating a user request and a service preference of an internet of vehicles, and belongs to the technical field of the internet of vehicles.
Background
With the development of information technology and the increase of social demands, the amount of wireless communication and information exchange between vehicles-X (X: vehicles, roads, pedestrians, the internet and the like) in the internet of vehicles is increased rapidly, and how to efficiently transmit data in the internet of vehicles system becomes a problem to be solved urgently.
The collaborative filtering recommendation algorithm based on clustering is known for its superior algorithm performance and higher accuracy. The clustering algorithm comprises hierarchical clustering, partition type clustering, density-based clustering, model-based clustering, spectral clustering and the like. Wherein, the hierarchical clustering mainly comprises top-down clustering and bottom-up clustering; the division type clustering mainly comprises k-means, k-means + + and the like; model-based clustering mainly includes Gaussian Mixture Models (GMMs) and the like; the density-based clustering algorithm mainly comprises DBSCAN, OPTICS and the like.
However, the conventional clustering recommendation can obtain good recommendation performance on a small-scale data set, but cannot adapt to a vehicle networking system with mass data, and mostly only considers users and services unilaterally, ignores the correlation between the users and the services, and causes lower recommendation result accuracy.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, provides a recommendation method and a recommendation system integrating user requests and service preferences in the Internet of vehicles, and can improve the accuracy of user service recommendation results in the Internet of vehicles.
In order to achieve the purpose, the invention is realized by adopting the following technical scheme:
in a first aspect, the invention provides a recommendation method for integrating a vehicle networking user request and service preference, which comprises the following steps:
receiving a service request data set for a vehicle, the service request data set for the vehicle including service requirements and vehicle context information; the vehicle context information comprises positioning information, driving speed, driving direction and residual oil quantity of the vehicle;
and generating recommendation information according to the service request data set and outputting and displaying the recommendation information.
Further, the method for generating recommendation information according to the service request data set and outputting and displaying the recommendation information comprises an online recommendation method;
the online recommendation method comprises the following steps:
receiving, by a network layer, a service request data set of a vehicle issued by the terminal layer, the service request data set of the vehicle including service requirements and vehicle context information;
extracting and preprocessing service data according to the service request data set to obtain a service data alternative set;
and executing a bidirectional spectral clustering algorithm on the service data alternative set to generate an online recommendation strategy and outputting and displaying the online recommendation strategy.
Further, the performing service data extraction and preprocessing comprises the following steps:
calculating the maximum driving mileage of the vehicle according to the vehicle context information;
a service data set within a maximum range is obtained based on the maximum range of the vehicle.
Further, the method for generating the online recommendation strategy and outputting the display by executing the bidirectional spectral clustering algorithm on the service data alternative set comprises the following steps:
obtaining a matrix S = { S } for each service in the service data candidate set i,j } mn Wherein s is i,j A value of credit to service j for user i;
carrying out normalization processing on the matrix of each service to finally obtain a normalization matrix W;
constructing the following diagonal matrix based on the normalized matrix W:
for a given feature matrix W, calculate
Wherein, A 1 And A 2 Is a diagonal matrix of a normalized matrix W, W n Is a laplacian matrix.
For a given feature matrix W n Calculate the first q = [ lbk ] of W]Left and right singular vectors L corresponding to singular values not equal to 1 t ＝(l 1 ,l 2 ,…,l q+1 ) And R t ＝(r 1 ,r 2 ,…r q+1 ) Wherein L is t Left singular vector characterization clustering result of row dimension, R t The right singular vector represents the clustering effect on the column dimensionality, and a feature vector space M after dimensionality reduction is generated:
clustering M by using a one-dimensional K-means algorithm; selecting a subset of the optimal left and right singular vectors, projecting data to the optimal subset of the singular vectors and clustering to obtain categories corresponding to users and services;
selecting a nearest neighbor set of a target service; for the services in different clustering sets, selecting the most similar service according to similarity calculation;
obtaining the prediction score P of the user u to the target service x through the score of u to all items in N (x) u,x And N (x) is the nearest neighbor set of the target service x, and the calculation formula is as follows:
wherein, P u,x For predictive scoring, sim (x, y) represents the similarity of the target service x to the nearest neighbor y, R u,y Represents the user u's score for service y,
mean scores for services x and y, respectively;
and calculating the prediction scores of all the unscored services of the target user u, sorting the unscored services from large to small according to the prediction scores, selecting the top N services with the highest scores as recommendation strategies to be issued to the road side unit, and sending recommendation service information to the vehicle through the communication of the road side unit.
Further, the method for selecting the most similar service according to the similarity calculation comprises the following steps: calculating the similarity between services through an Euclidean distance formula, and selecting the first k services with the highest similarity with the services as nearest neighbor sets;
where Sim (x, y) is the similarity of services x and y, x i And y i Coordinates of two services respectively, n represents a dimension; the smaller the distance is, the two clothesThe more similar the service, the greater the similarity, and the more in line with the user preference.
Further, the method for generating recommendation information according to the service request data set and outputting and displaying also comprises an off-line recommendation method;
the off-line recommendation method comprises the following steps:
obtaining historical scoring data of a user, obtaining historical scoring data of the user, taking the historical scoring data of the user as a data set and storing the historical scoring data into the historical scoring data, and calculating the similarity between services according to the following formula:
where Sim (x, y) is the similarity of services x and y, x i And y i Coordinates of two services are respectively represented, and n represents a dimension; the smaller the distance is, the more similar the two services are, the greater the similarity is, and the more the services meet the preference requirements of the user;
introducing a service distance influence factor and a scoring factor when calculating user preference service, wherein the calculation formula is as follows:
wherein r represents a service prediction score, α, β, and γ are influence factors of the similarity, distance, and service score, respectively, and α + β + γ =1;
and sorting according to the r value, and selecting the service with the top rank as a recommendation set.
In a second aspect, the invention provides a recommendation system integrating a user request and a service preference of an internet of vehicles, which comprises a terminal layer, a network layer, a cloud platform layer and an application layer;
the terminal layer is used for acquiring vehicle context information; the vehicle context information comprises positioning information, driving speed, driving direction and residual oil quantity of the vehicle;
the network layer is used for information exchange between the terminal layer and the cloud platform layer;
the cloud platform is used for generating recommendation information according to the vehicle context information and the service requirements and transmitting the recommendation information to the application layer;
the application layer is used for interaction between a user and an application program and displaying recommendation information, and comprises a plurality of visual application software programs.
Further, the cloud platform layer comprises an online recommendation device and an offline recommendation device;
the online recommendation device comprises:
an information receiving module: the terminal layer is used for receiving a service request data set of the vehicle sent by the terminal layer, and the service request data set of the vehicle comprises service requirements and vehicle context information;
the information extraction module: the service data backup and preprocessing module is used for extracting and preprocessing service data according to the service request data set to obtain a service data alternative set;
a service recommendation module: and the bidirectional spectral clustering algorithm is used for executing the bidirectional spectral clustering algorithm on the preprocessed data to generate an online recommendation strategy and outputting and displaying the online recommendation strategy.
Further, the service recommendation module is configured to perform the following steps:
obtaining a matrix S = { S } for each service in the service data alternative set i,j } mn Wherein s is i,j A value of credit to service j for user i;
carrying out normalization processing on the matrix of each service to finally obtain a normalization matrix W;
constructing the following diagonal matrix based on the normalized matrix W:
for a given feature matrix W, calculate
For a given feature matrix W, calculate the first q = [ lbk ] of W]Left and right singular vectors L corresponding to singular values not equal to 1 t ＝(l 1 ,l 2 ,…,l q+1 ) And R t ＝(r 1 ,r 2 ,…r q+1 ) Wherein, L t Left singular vector characterization clustering result of line dimensions, R t The right singular vector represents the clustering effect on the column dimensionality and generates a feature vector space after dimensionality reduction:
clustering the M by using a one-dimensional K-means algorithm; selecting a subset of the optimal left and right singular vectors, projecting data to the optimal subset of the singular vectors and clustering to obtain categories corresponding to users and services;
selecting a nearest neighbor set of a target service; for the services in different clustering sets, selecting the most similar service according to similarity calculation;
and obtaining the predicted score of the user u on the target service x through the score of u on all items in N (x), wherein N (x) is the nearest neighbor set of the target service x, and the calculation formula is as follows:
where Sim (x, y) represents the similarity of the target service x to the nearest neighbor y, R u,y Represents the user u's score for service y,
mean scores for services x and y, respectively;
and calculating the prediction scores of all the unscored services of the target user u, sorting the unscored services from large to small according to the prediction scores, selecting the first N services with the highest scores as recommendation strategies, issuing the recommendation strategies to the road side unit, and sending recommendation service information to the vehicle through communication of the road side unit.
Further, the cloud platform layer further comprises an offline recommendation device;
the offline recommendation device comprises:
a similarity module: the system is used for acquiring historical scoring data of users, storing the historical scoring data of the users as a data set in the historical data, and calculating the similarity between services according to the following formula:
where Sim (x, y) represents the similarity of the target service x to the nearest neighbor y, x i And y i Coordinates of two services respectively, n represents a dimension; the smaller the distance is, the more similar the two services are, the greater the similarity is, and the more the services meet the preference requirements of users;
a score calculating module: the method is used for introducing a service distance influence factor and a scoring factor when calculating the user preference service, and the calculation formula is as follows:
wherein r represents service prediction score, α, β and γ are influence factors of similarity, distance and service score, respectively, and α + β + γ =1;
a sorting recommendation module: and the system is used for sorting according to the value of the score r and selecting the service with the top rank as a recommendation set.
Compared with the prior art, the invention has the following beneficial effects:
1. according to the invention, the service data set is filtered through the cost function of the internet vehicle to obtain a reasonable personalized data set, so that the problem that the traditional recommendation system does not consider personalization and vehicle running context at the same time is solved, the calculation time consumption is reduced, and the time delay is reduced;
2. a Collaborative Filtering recommendation algorithm (CFBS) based on user and project Bidirectional spectrum clustering is adopted in the online recommendation of the Internet of vehicles, users and services are clustered in a Bidirectional spectrum mode, user classes and service classes are integrated, a nearest neighbor set is searched, the similarity of the classes between the users and the classes between the services is considered, the cohesion is stronger, and the discrimination between the classes is larger. Therefore, the recommendation quality is higher than that of a unilateral collaborative filtering algorithm based on service clustering;
3. the information issuing service is provided for the online vehicle through two stages of online recommendation and offline training. The online recommendation method comprises the steps that online vehicle service requirements are acquired in real time in an online recommendation stage, server information is efficiently issued to online vehicle users through a collaborative filtering recommendation algorithm based on user requirements and service bidirectional spectral clustering, and the server information is synchronously stored in user scoring records; and in the off-line stage, the user scoring data is acquired and fused, calculation is carried out through a user preference model, and the issued information is synchronously updated.
Drawings
FIG. 1 is a service recommendation system framework diagram;
FIG. 2 is a flow chart of a service recommendation method;
FIG. 3 is a comparison graph of average distance traveled for different parameter settings
FIG. 4 is a comparison graph of service recommendation success rates at different parameter settings;
FIG. 5 is a comparison graph of recommendation policy service recommendation success rates.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
The first embodiment is as follows:
the embodiment provides a recommendation method for integrating a vehicle networking user request and service preference, which comprises the following steps of:
receiving a service request data set for a vehicle, the service request data set for the vehicle including service requirements and vehicle context information; the vehicle context information comprises positioning information, driving speed, driving direction and residual oil quantity of the vehicle;
and generating recommendation information according to the service request data set and outputting and displaying the recommendation information.
Further, the method for generating recommendation information according to the service request data set and outputting and displaying the recommendation information comprises an online recommendation method;
the online recommendation method comprises the following steps:
receiving, by a network layer, a service request data set of a vehicle sent by the terminal layer, the service request data set of the vehicle including a service requirement and vehicle context information;
extracting and preprocessing service data according to the service request data set to obtain a service data alternative set;
and executing a bidirectional spectral clustering algorithm on the service data alternative set to generate an online recommendation strategy.
Specifically, the service data extraction and preprocessing include the following steps:
calculating the maximum driving mileage of the vehicle according to the vehicle context information;
a service data set within a maximum range is obtained based on the maximum range of the vehicle.
Specifically, the performing a bidirectional spectral clustering algorithm on the service data alternative set to generate an online recommendation strategy includes:
obtaining a matrix S = { S } for each service in the service data candidate set i,j } mn Wherein s is i,j User i to service j;
carrying out normalization processing on the matrix of each service to finally obtain a normalization matrix W;
constructing the following diagonal matrix based on the normalized matrix W:
for a given feature matrix W, calculate
For a given feature matrix W n Calculating W n Left and right singular vectors L corresponding to the first k singular values of not 1 t ＝(l 1 ,l 2 ,…,l q+1 ) And R t ＝(r 1 ,r 2 ,…r q+1 ) Wherein L is t Left singular vector characterization clustering result of row dimension, R t The right singular vector represents the clustering effect on the column dimensionality, and a feature vector space after dimensionality reduction is generated:
clustering the M by using a one-dimensional K-means algorithm; selecting a subset of the optimal left and right singular vectors, projecting data to the optimal subset of the singular vectors and clustering to obtain categories corresponding to users and services;
selecting a nearest neighbor set of a target service; for the services in different clustering sets, selecting the most similar service according to similarity calculation; calculating the similarity between services through an Euclidean distance formula, and selecting the first k services with the highest similarity with the services as nearest neighbor sets;
where Sim (x, y) represents the similarity of the target service x to the nearest neighbor y, x i And y i Coordinates of two services respectively, n represents a dimension; the smaller the distance is, the more similar the two services are, the greater the similarity is, and the more the services meet the preference requirements of the user;
and obtaining the predicted score of the user u on the target service x through the score of u on all items in N (x), wherein N (x) is the nearest neighbor set of the target service x, and the calculation formula is as follows:
where Sim (x, y) represents the similarity of the target service x to the nearest neighbor y, R u,y Represents the user u's score for service y,
mean scores for services x and y, respectively;
and calculating the prediction scores of all the unscored services of the target user u, sorting the unscored services from large to small according to the prediction scores, selecting the top N services with the highest scores as recommendation strategies to be issued to the road side unit, and sending recommendation service information to the vehicle through the communication of the road side unit.
Specifically, the offline recommendation method comprises the following steps:
obtaining historical scoring data of the user, storing the historical scoring data of the user as a data set in the historical data, and calculating the similarity between services by using the following formula:
where Sim (x, y) represents the similarity of the target service x to the nearest neighbor y, x i And y i Coordinates of two services respectively, n represents a dimension; the smaller the distance is, the more similar the two services are, the greater the similarity is, and the more the services meet the preference requirements of users;
introducing a service distance influence factor and a scoring factor when calculating user preference service, wherein the calculation formula is as follows:
wherein, α, β and γ are influence factors of similarity, distance and service score, respectively, and α + β + γ =1;
and sorting according to the r value, and selecting the service with the top rank as a recommendation set.
Example two:
the embodiment provides a recommendation system integrating a user request and a service preference in the internet of vehicles, which comprises a terminal layer, a network layer, a cloud platform layer and an application layer, as shown in fig. 1;
the terminal layer is used for acquiring vehicle context information; the vehicle context information comprises positioning information, driving speed, driving direction and residual oil quantity of the vehicle; the terminal layer obtains vehicle positioning through a global positioning system GPS, and a large amount of vehicle context information such as the driving speed, the driving direction, the residual oil quantity and the like of the current vehicle can be obtained through various sensor devices of the vehicle. The context information and the hardware equipment lay a foundation for realizing the upper layer function; the network layer can realize communication among the in-vehicle components and remote data transmission, the communication of the vehicle components mainly refers to the transmission of data information acquired by the equipment layer to the cloud platform, and the remote transmission mainly refers to the communication among the vehicles and the communication of Road Side Units (RSUs) and comprises data transmission and service requests; the cloud platform is used as a bridge for connecting a network and application services and is responsible for calculation of a recommendation algorithm and information release services; the application layer provides a plurality of visual application software programs for vehicle users, and the vehicle users can operate through a visual interface to realize human-computer interaction.
The terminal layer is also an internet vehicle user layer and is mainly responsible for acquiring a large amount of vehicle context information. Taking the parking lot push service as an example, suppose that the set of networked vehicles is C = { C = } 1 ,c 2 ,…,c n }, vehicle c k Is expressed as (x) k ,y k ) (k =1,2, \8230;, n), vehicle c k Is at a destination position of (x) d ,y d ). The contextual information of the vehicle includes: the method comprises the steps that a networked vehicle acquires GPS position information through a satellite navigation system; vehicle c k The OBU (on-board unit) terminal is obtained through an on-board OBD interface (or CAN bus)And obtaining vehicle working condition information.
The network layer is used for information communication between the terminal layer and the cloud platform layer; the network layer is mainly responsible for communication among all components in the vehicle and remote data communication, the communication of all the components in the vehicle mainly means that context information acquired by the bottom layer equipment is transmitted to the application layer through a network, and the remote transmission mainly comprises communication among vehicles and communication between the vehicles and the RSU.
The cloud platform is used for generating recommendation information according to the vehicle context information and the service requirements and transmitting the recommendation information to the application layer; and the cloud platform layer is responsible for calculation of the recommendation algorithm and information release service. The service recommendation flow of the cloud platform layer is shown in fig. 2. The method mainly comprises two parts of online recommendation and offline recommendation.
The online recommendation device is characterized in that firstly, a networked vehicle sends out a service request, the service request carries the current vehicle position and context information (vehicle working condition information, neighbor vehicle information and the like), then the service request is reported to a cloud platform layer through an RSU (remote server unit), a data set is filtered through an information extraction module, and then a bidirectional spectral clustering algorithm is executed by a service recommendation module to generate an online recommendation strategy (the online recommendation strategy refers to real-time recommendation information of the vehicle, such as real-time parking lot information near a destination). The specific processing steps are as follows.
The information extraction module mainly provides a cost function scheme to realize the preprocessing of the service data set, namely, the service data set reported by the network layer is extracted and filtered, so that the precision of the data set used in the recommendation process is improved. The specific steps are as follows.
Step 1.1: and reflecting the maximum driving mileage of the vehicle through the cost function of the networked vehicles when the recommended strategy is calculated. The cost function of the networked vehicle includes the remaining fuel amount, or remaining power amount, or remaining balance after charging, etc., and examples are shown in table 1.
TABLE 1 cost function for networked vehicles
Wherein the content of the first and second substances,
o represents the amount of the remaining oil,
represents average fuel consumption; e is the remaining energy of the battery,
e is the amount of energy change over a statistical time,
d is the driving distance in the statistical time; p is the gasoline unit price.
Setting the coordinate position of the parking lot service as (x) s ,y s ) The distance from the destination to each service is
According to the cost function of the networked vehicles, the reachable travel distance of the vehicles is given as
Wherein F is a cost function of the networked vehicle, q and g are vehicle constant parameters, v is the running speed of the vehicle, and L is the traction of the vehicle. The method comprises the steps that a requesting vehicle obtains a vehicle surrounding service data set through RSU communication, the distance dist between the requesting vehicle and a service is calculated through a distance formula (9), the maximum driving mileage Mdist which can be reached by the requesting vehicle is calculated through a formula (10), and if the dist is smaller than Mdist, the position of the service is within the reach range of the requesting vehicle, and the service is added to the service data set. The pseudo code for the selected data set is shown in table 2.
Table 2 data set filtering selection procedure
And 1.2, extracting the service data set screened in the maximum distance range based on the step 1.1. A user set U and a service set I are obtained, and a user-service scoring matrix S is constructed as shown in table 3, and generally, an mxn-order matrix can be used to represent scoring data of a user on a service. Where m denotes the number of users and n denotes the number of services.
TABLE 3 user-service matrix
The service recommendation module is mainly used for generating an online recommendation strategy (the online recommendation strategy refers to real-time recommendation information of the vehicle, such as real-time parking lot information near a destination) based on a Collaborative Filtering recommendation algorithm (CFBS) for Bidirectional Spectral clustering of users and items. The specific treatment steps are as follows.
Step 2.1: acquiring user service matrix S = { S = } i,j } mn Wherein s is i,j The value of the credit to service j is given to user i. The normalization processing is carried out, and the processing steps are as follows.
Step 2.1.2-first calculate the logarithm:
L ij ＝log(S) (11)
step 2.1.3 calculate the average of the ith row:
step 2.1.4-calculate average value in column j:
step 2.1.5, calculating the average value of the whole matrix:
step 2.1.6, calculating the final normalization matrix as follows:
W＝(W ij ) (16)
finally, a normalized matrix W is obtained.
Step 2.2, constructing the following diagonal matrix based on the characteristic matrix W:
step 2.3-for a given feature matrix W, calculate
For a given feature matrix W n Calculating W n The left and right singular vectors corresponding to the first k singular values not equal to 1 are respectively L t ＝(l 1 ,l 2 ,…,l k+1 ) And R t ＝(r 1 ,r 2 ,…r k+1 ) Wherein L is t Left singular vector characterization clustering result of line dimensions, R t The right singular vector characterizes the clustering effect on the column dimensions. Generating a feature vector space after dimension reduction:
step 2.4: m is clustered using a one-dimensional K-means algorithm. A subset of the best left and right singular vectors is selected and the data is projected onto the best subset of singular vectors and clustered. And obtaining the corresponding categories of the user and the service.
Step 2.5: a nearest neighbor set of the target service is selected. And for the services in different clustering sets, selecting the most similar service according to similarity calculation, wherein the specific formula is as follows, the similarity between the services is calculated by using a Euclidean distance formula, and the first k services with the highest similarity to the service are selected as the nearest neighbor set.
Wherein x is i And y i The coordinates of the two services are respectively, and n represents the dimension. The smaller the distance, the more similar the two services are, the greater the similarity is, the more in line with the user preference requirements.
And 2.6, setting the nearest neighbor set of the target service x as N (x), and predicting the score of the user on the target service according to the score information of the user on the neighbor. The predicted score of the user u for the target service x can be obtained by scoring all items in N (x) by u, and the calculation formula is as follows.
Where Sim (x, y) represents the similarity of the target service x to the nearest neighbor y, R u,y Represents the rating of service y by user u,
mean scores for services x and y are represented, respectively.
And 2.7, calculating the prediction scores of all unscored services of the target user u according to the method, sequencing the items from large to small according to the prediction scores, and selecting the top N services with the highest scores. And issuing the recommendation strategy to the RSU, and acquiring recommendation service information by the networked vehicle through communication with the RSU.
The off-line recommendation part is mainly used for calculating recommendation service through a user preference model and synchronizing the recommendation service to a recommendation service module to update an off-line recommendation strategy. The specific process comprises the following steps.
And 3.1, acquiring historical scoring data of the user, storing the historical scoring data of the user as a data set in the historical data, and calculating the similarity between services by using an equation (14).
And 3.2, in order to ensure that the recommended service does not influence the driving route of the vehicle as much as possible and ensure the service quality. The service distance influence factor and the scoring factor are introduced when the user preference service is calculated, and the calculation formula is as follows.
Wherein, α, β and γ are influence factors of similarity, distance and service score, respectively, α + β + γ =1, and if α is higher, it indicates that the service recommended to the user emphasizes the service preference of the user, but excessive pursuit of the service preference may bring a detour behavior and further increase the travel distance, the influence of the service score becomes smaller, and the recommended service quality is also influenced. If beta is higher, it indicates that the service recommended to the user is more focused on the distance of the service, but excessive pursuit of the distance may bring low user preference and service quality problems. If gamma is higher, it indicates that the service recommended to the user is a score that places more emphasis on the service, which also corresponds to the quality of service. Excessive pursuit of quality of service may present travel distance and user preference issues. Therefore, the user can adjust the weight of β and γ according to the actual situation, for example, the current vehicle is full, and only wants to find a high-quality service according to his preference. In general, the balance of α, β and γ is taken. And finally, sorting according to the r value, and selecting the first 5 services as a recommendation set.
The application layer is used for interacting with a user through application programs, displaying recommendation information and acquiring service requirements, and comprises a plurality of visual application software programs. The application layer is responsible for interaction between the user and the application program, issues the recommendation strategy generated by the network layer to the OBU of the networked vehicle, and stores the evaluation of the user on the request service into the historical record database for calculating the user preference model.
Taking a parking lot service scene as an example, the effects of the 4 service selection strategies are compared through a set of simulation experiments, and the experimental results are analyzed and compared. The 4 service selection policies are a shortest distance selection policy (SDS), a composite score selection policy (ARS), a context based service recommendation policy (CBSR), and a recommendation policy proposed herein, respectively. The experimental data come from the Yelp review website, which includes 50 users who visited the simulation area, which ranged from 4000m × 4000m.
Step 4.1: specific experimental simulation parameter settings are shown in table 4.
Table 4 experimental simulation parameter settings
Step 4.2: the service recommendation quality evaluation indexes of the experiment comprise service recommendation success rate and average driving distance. The expression of the service recommendation success rate is shown below.
The service success recommendation rate represents that the service satisfied by the user accounts for the proportion of all recommended services.
The average travel distance of the service represents the average distance the user needs to travel to reach the service location according to the recommended strategy, and the specific formula is as follows.
Wherein, distance is the distance from the user to the service provided each time, and number is the total number of the service recommended each time.
Step 4.3: experimental results and analysis. The recommendation selection of the experiment is based on the calculation result of the recommendation coefficient of formula 16, and different parameter designs have direct influence on the final recommendation result. FIG. 3 shows s when α takes different values between 0 and 1 m And p s . In this case, when γ =0.2 and α is large, p is set to be larger as shown in fig. 3 s Is high, but s m Is also longer, with increasing alpha, p s The rate of increase of (a) is also decreasing, and s m The growth rate of (c) is increasing. Therefore to guarantee p s While minimizing s m Thus, α =0.6 and β =0.2 are selected.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.
Claims (6)
1. A recommendation method fusing a vehicle networking user request and a service preference is characterized by comprising the following steps:
receiving a service request data set for a vehicle, the service request data set for the vehicle including service requirements and vehicle context information; the vehicle context information comprises positioning information, driving speed, driving direction and residual oil quantity of the vehicle;
generating recommendation information according to the service request data set and outputting and displaying the recommendation information;
the method for generating recommendation information according to the service request data set and outputting and displaying the recommendation information comprises an online recommendation method;
the online recommendation method comprises the following steps:
receiving a service request data set of a vehicle sent by a terminal layer through a network layer, wherein the service request data set of the vehicle comprises service requirements and vehicle context information;
extracting and preprocessing service data according to the service request data set to obtain a service data alternative set;
executing a bidirectional spectral clustering algorithm on the service data alternative set to generate an online recommendation strategy and outputting and displaying the online recommendation strategy;
the method for executing the bidirectional spectral clustering algorithm on the service data alternative set to generate the online recommendation strategy and outputting and displaying comprises the following steps:
obtaining a matrix S = { S } for each service in the service data candidate set i,j } mn Wherein s is i,j For user i to service j;
carrying out normalization processing on the matrix of each service to finally obtain a normalization matrix W;
constructing the following diagonal matrix based on the normalized matrix W:
for a given feature matrix W, calculate
Wherein A is 1 And A 2 Is a diagonal matrix of a normalized matrix W, W n Is a Laplace matrix;
for a given feature matrix W n Calculating left and right singular vectors L corresponding to the first k singular values of W which are not 1 t ＝(l 1 ,l 2 ,…,l q+1 ) And R t ＝(r 1 ,r 2 ,…r q+1 ) Wherein L is t Left singular vector characterization clustering result of row dimension, R t The right singular vector represents the clustering effect on the column dimensionality, and a feature vector space M after dimensionality reduction is generated:
clustering M by using a one-dimensional K-means algorithm; selecting a subset of the optimal left and right singular vectors, projecting data to the optimal subset of the singular vectors and clustering to obtain categories corresponding to users and services;
selecting a nearest neighbor set of a target service; for the services in different clustering sets, selecting the most similar service according to similarity calculation;
obtaining the prediction score P of the user u to the target service x through the score of u to all items in N (x) u,x And N (x) is the nearest neighbor set of the target service x, and the calculation formula is as follows:
wherein, P u,x For predictive scoring, sim (x, y) represents the similarity of the target service x to the nearest neighbor y, R u,y Represents the rating of service y by user u,
mean scores for services x and y, respectively;
and calculating the prediction scores of all the unscored services of the target user u, sorting the unscored services from large to small according to the prediction scores, selecting the top N services with the highest scores as recommendation strategies to be issued to the road side unit, and sending recommendation service information to the vehicle through the communication of the road side unit.
2. The recommendation method for fusion of user request and service preference in the internet of vehicles according to claim 1, wherein the service data extraction and pre-processing comprises the following steps:
calculating the maximum driving mileage of the vehicle according to the vehicle context information;
a service data set within a maximum range is obtained based on the maximum range of the vehicle.
3. The recommendation method fusing the car networking user request and the service preference according to claim 1, wherein the method of selecting the most similar service for the car networking user request according to the similarity calculation comprises the following steps: calculating the similarity between services through an Euclidean distance formula, and selecting the first k services with the highest similarity with the services as nearest neighbor sets;
wherein Sim (x, y) is the similarity of service x and y, x i And y i Coordinates of two services respectively, n represents a dimension; the smaller the distance, the more similar the two services are, the greater the similarity is, the more in line with the user preference requirements.
4. The recommendation method fusing the Internet of vehicles user request and the service preference according to claim 1, wherein the method for generating recommendation information according to the service request data set and outputting the recommendation information for display further comprises an offline recommendation method;
the off-line recommendation method comprises the following steps:
obtaining historical scoring data of the user, storing the historical scoring data of the user as a data set in the historical data, and calculating the similarity between services by using the following formula:
where Sim (x, y) is the similarity of services x and y, x i And y i Coordinates respectively representing two services, and n represents a dimension; the smaller the distance is, the more similar the two services are, the greater the similarity is, and the more the services meet the preference requirements of users;
introducing a service distance influence factor and a scoring factor when calculating the user preference service, wherein the calculation formula is as follows:
wherein r represents a service prediction score, α, β, and γ are influence factors of the similarity, distance, and service score, respectively, and α + β + γ =1; sim is similarity, dist is the distance from the destination reached by the networked vehicle to each service, mdist is the reachable driving mileage of the vehicle, and s is service score;
and sorting according to the r value, and selecting the service with the top rank as a recommendation set.
5. A recommendation system integrating a user request and service preference of the Internet of vehicles is characterized by comprising a terminal layer, a network layer, a cloud platform layer and an application layer;
the terminal layer is used for acquiring vehicle context information; the vehicle context information comprises positioning information, driving speed, driving direction and residual oil quantity of the vehicle;
the network layer is used for information exchange between the terminal layer and the cloud platform layer;
the cloud platform is used for generating recommendation information according to the vehicle context information and the service requirements and transmitting the recommendation information to the application layer;
the application layer is used for interaction between a user and an application program and displaying recommendation information, including various visual application software programs;
the cloud platform layer comprises an online recommendation device and an offline recommendation device;
the online recommendation device comprises:
an information receiving module: the terminal layer is used for receiving a service request data set of the vehicle sent by the terminal layer, and the service request data set of the vehicle comprises service requirements and vehicle context information;
the information extraction module: the service data backup and preprocessing module is used for extracting and preprocessing service data according to the service request data set to obtain a service data alternative set;
the service recommendation module: the bidirectional spectral clustering algorithm is used for executing the bidirectional spectral clustering algorithm on the preprocessed data to generate an online recommendation strategy and outputting and displaying the online recommendation strategy;
the service recommendation module is used for executing the following steps:
obtaining a matrix S = { S } for each service in the service data alternative set i,j } mn Wherein s is i,j A value of credit to service j for user i;
carrying out normalization processing on the matrix of each service to finally obtain a normalization matrix W;
constructing the following diagonal matrix based on the normalized matrix W:
for a given feature matrix W, calculate
For a givenCalculating a characteristic matrix W, wherein left and right singular vectors corresponding to first k singular values which are not 1 of the characteristic matrix W are respectively L t ＝(l 1 ,l 2 ,…,l q+1 ) And R t ＝(r 1 ,r 2 ,…r q+1 ) Wherein L is t Left singular vector characterization clustering result of row dimension, R t The right singular vector represents the clustering effect on the column dimensionality and generates a feature vector space after dimensionality reduction:
clustering the M by using a one-dimensional K-means algorithm; selecting a subset of the optimal left and right singular vectors, projecting data to the optimal subset of the singular vectors, clustering, and obtaining categories corresponding to users and services;
selecting a nearest neighbor set of a target service; for the services in different clustering sets, selecting the most similar service according to similarity calculation;
and obtaining the prediction score of the user u on the target service x through the score of u on all items in N (x), wherein N (x) is the nearest neighbor set of the target service x, and the calculation formula is as follows:
where Sim (x, y) represents the similarity of the target service x to the nearest neighbor y, R u,y Represents the user u's score for service y,
mean scores for services x and y, respectively;
and calculating the prediction scores of all the unscored services of the target user u, sorting the unscored services from large to small according to the prediction scores, selecting the first N services with the highest scores as recommendation strategies, issuing the recommendation strategies to the road side unit, and sending recommendation service information to the vehicle through communication of the road side unit.
6. The recommendation system fusing vehicle networking user requests and service preferences according to claim 5, wherein the cloud platform layer further comprises an offline recommendation device;
the offline recommendation device comprises:
a similarity module: the system is used for acquiring historical scoring data of the user, storing the historical scoring data of the user as a data set in the historical data, and calculating the similarity between services according to the following formula:
where Sim (x, y) represents the similarity of the target service x to the nearest neighbor y, x i And y i Coordinates of two services respectively, n represents a dimension; the smaller the distance is, the more similar the two services are, the greater the similarity is, and the more the services meet the preference requirements of the user;
a score calculating module: the method is used for introducing a service distance influence factor and a scoring factor when calculating the user preference service, and the calculation formula is as follows:
wherein, α, β and γ are influence factors of similarity, distance and service score, respectively, and α + β + γ =1; sim is similarity, dist is the distance from the destination reached by the networked vehicle to each service, mdist is the reachable driving mileage of the vehicle, and s is service score;
a sorting recommendation module: and the system is used for sorting according to the value of the score r and selecting the service with the top rank as a recommendation set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110290783.1A CN112989194B (en) | 2021-03-18 | 2021-03-18 | Recommendation method and system integrating user request and service preference of Internet of vehicles |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110290783.1A CN112989194B (en) | 2021-03-18 | 2021-03-18 | Recommendation method and system integrating user request and service preference of Internet of vehicles |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112989194A CN112989194A (en) | 2021-06-18 |
| CN112989194B true CN112989194B (en) | 2022-11-29 |
Family
ID=76333116
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110290783.1A Active CN112989194B (en) | 2021-03-18 | 2021-03-18 | Recommendation method and system integrating user request and service preference of Internet of vehicles |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112989194B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113778556B (en) * | 2021-09-14 | 2023-09-19 | 湖南大学 | Task unloading recommendation method and system based on vehicle-road cooperation |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109471982A (en) * | 2018-11-21 | 2019-03-15 | 南京邮电大学 | A kind of web service recommendation method based on user and service cluster QoS perception |
| US10475105B1 (en) * | 2018-07-13 | 2019-11-12 | Capital One Services, Llc | Systems and methods for providing improved recommendations |
| CN111949891A (en) * | 2020-10-09 | 2020-11-17 | 广州斯沃德科技有限公司 | Personalized information recommendation method and system based on vehicle track clustering |
| CN111951070A (en) * | 2020-07-31 | 2020-11-17 | 上海博泰悦臻电子设备制造有限公司 | Intelligent recommendation method and device based on Internet of vehicles, server and storage medium |
-
2021
- 2021-03-18 CN CN202110290783.1A patent/CN112989194B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10475105B1 (en) * | 2018-07-13 | 2019-11-12 | Capital One Services, Llc | Systems and methods for providing improved recommendations |
| CN109471982A (en) * | 2018-11-21 | 2019-03-15 | 南京邮电大学 | A kind of web service recommendation method based on user and service cluster QoS perception |
| CN111951070A (en) * | 2020-07-31 | 2020-11-17 | 上海博泰悦臻电子设备制造有限公司 | Intelligent recommendation method and device based on Internet of vehicles, server and storage medium |
| CN111949891A (en) * | 2020-10-09 | 2020-11-17 | 广州斯沃德科技有限公司 | Personalized information recommendation method and system based on vehicle track clustering |
Non-Patent Citations (1)
| Title |
|---|
| 互联网汽车出行娱乐系统需求分析及系统设计;罗彭沪京;《中国优秀硕士学位论文全文数据库信息科技辑》;20190115;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112989194A (en) | 2021-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11842404B2 (en) | Enhancement using analytics based on vehicle kinematic data | |
| US11586992B2 (en) | Travel plan recommendation method, apparatus, device and computer readable storage medium | |
| US20220187087A1 (en) | Systems and methods for predicting fuel consumption efficiency | |
| Bader et al. | Context-aware POI recommendations in an automotive scenario using multi-criteria decision making methods | |
| WO2021136476A1 (en) | System and method for alternative destination recommendation on ridesharing platforms | |
| CN105282224B (en) | The method and apparatus for determining opposite driving performance using vehicle participatory sensing system | |
| CN112766607A (en) | Travel route recommendation method and device, electronic device and readable storage medium | |
| US20200210905A1 (en) | Systems and Methods for Managing Networked Vehicle Resources | |
| CN111428137B (en) | Recommendation method and recommendation device for electric vehicle charging facilities | |
| CN105678412A (en) | Path planning method and device facing multiple passengers | |
| CN111859178B (en) | Method and system for recommending get-on point | |
| CN111127130A (en) | Energy site recommendation method based on user preference, storage medium and electronic equipment | |
| CN111523932A (en) | Scoring method, device and system for network car booking service and storage medium | |
| CN111861643A (en) | Riding position recommendation method and device, electronic equipment and storage medium | |
| US20230273040A1 (en) | System and method of creating custom dynamic neighborhoods for individual drivers | |
| CN111932341A (en) | Method and system for determining car pooling order | |
| CN111400425B (en) | Method and system for automatically optimizing and selecting paths | |
| CN112989194B (en) | Recommendation method and system integrating user request and service preference of Internet of vehicles | |
| CN109767020B (en) | Vehicle recommendation method, device, computer equipment and storage medium | |
| WO2024104012A1 (en) | Lane positioning method and apparatus, and computer device, computer-readable storage medium and computer program product | |
| CN112070377B (en) | Travel service processing method and device, electronic equipment and storage medium | |
| CN116611891A (en) | Content information recommendation method, device, server and storage medium | |
| CN112102134A (en) | Event processing method, system, device, computer readable storage medium and equipment | |
| Leurent et al. | Smart mobility: A landscape under development | |
| US12033091B2 (en) | Matching drivers with shared vehicles to optimize shared vehicle services |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |