CN113032670B - Parking lot recommendation method and device, computer equipment and storage medium - Google Patents

Parking lot recommendation method and device, computer equipment and storage medium Download PDF

Info

Publication number
CN113032670B
CN113032670B CN202110276770.9A CN202110276770A CN113032670B CN 113032670 B CN113032670 B CN 113032670B CN 202110276770 A CN202110276770 A CN 202110276770A CN 113032670 B CN113032670 B CN 113032670B
Authority
CN
China
Prior art keywords
parking lot
model
matrix
scoring
updated
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
Application number
CN202110276770.9A
Other languages
Chinese (zh)
Other versions
CN113032670A (en
Inventor
范玉顺
卢巧渝
伍星
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tsinghua University
Original Assignee
Tsinghua University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tsinghua University filed Critical Tsinghua University
Priority to CN202110276770.9A priority Critical patent/CN113032670B/en
Publication of CN113032670A publication Critical patent/CN113032670A/en
Application granted granted Critical
Publication of CN113032670B publication Critical patent/CN113032670B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/602—Providing cryptographic facilities or services
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
    • G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
    • G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Databases & Information Systems (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Development Economics (AREA)
  • Health & Medical Sciences (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioethics (AREA)
  • Quality & Reliability (AREA)
  • Computer Hardware Design (AREA)
  • Software Systems (AREA)
  • Educational Administration (AREA)
  • Computer Security & Cryptography (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Data Mining & Analysis (AREA)
  • Medical Informatics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Traffic Control Systems (AREA)

Abstract

The application relates to a parking lot recommendation method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring a scoring matrix and a parking lot characteristic image matrix of the candidate parking lot; establishing an updating iterative model according to the user characteristic image matrix, the grading matrix and the parking lot characteristic image matrix, iterating the updating iterative model to obtain a model gradient result, and sending the model gradient result to the aggregation server; receiving a parking lot characteristic image matrix which is sent by an aggregation server and is updated according to a model gradient result; iteratively updating the updated iterative model according to the updated parking lot feature portrait matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value to obtain a final prediction score value of the candidate parking lot; and scoring each candidate parking lot model according to the final predicted scoring value to obtain a scoring result, and recommending parking lots to the target user according to the scoring result. By adopting the method, the personal privacy data security of the user can be ensured.

Description

Parking lot recommendation method and device, computer equipment and storage medium
Technical Field
The present application relates to the field of distributed data processing technologies, and in particular, to a parking lot recommendation method and apparatus, a computer device, and a storage medium.
Background
With the development of internet technology, more and more technologies need big data as support, however, a data privacy problem is derived for the source and content of data, and for personal privacy data provided by internet users, service providers cannot use the data without authorization, so that data held by data holders are not associated with each other and cannot be shared at will, and a data island problem is formed.
In a recommended scene of a parking lot, parking data of a user side may relate to privacy data such as license plate numbers, parking places, parking time and the like, and data of each user relative to the parking lot is like a seat data island and cannot be shared with a service providing side for recommending the parking lot according to user preference.
Disclosure of Invention
In view of the above, it is necessary to provide a parking lot recommendation method, apparatus, computer device and storage medium for solving the above technical problems.
A parking lot recommendation method, the method comprising:
acquiring a scoring matrix and a parking lot characteristic image matrix of a candidate parking lot; the scoring matrix comprises historical scores of each candidate parking lot by a target user;
establishing an updating iterative model for predicting scores according to the user characteristic portrait matrix, the scoring matrix and the parking lot characteristic portrait matrix, iterating the updating iterative model to obtain a model gradient result, and sending the model gradient result to an aggregation server;
receiving the updated parking lot feature image matrix sent by the aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to the model gradient result;
iteratively updating the updated iterative model according to the updated parking lot feature image matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value, and obtaining a final prediction score value of the candidate parking lot;
and scoring each candidate parking lot model according to the final prediction scoring value to obtain a scoring result, and recommending parking lots to the target user according to the scoring result.
In one embodiment, the obtaining a score matrix and a parking lot feature image matrix of the candidate parking lot includes:
acquiring historical parking lot data of target users, and counting in the historical parking lot data to obtain a parking time data record of each target user for each candidate parking lot;
carrying out normalization processing on the parking time data records of each candidate parking lot of each target user to obtain the score value of each target user on each candidate parking lot;
merging the score value of each target user to obtain a score matrix of each candidate parking lot;
and downloading the encrypted parking lot characteristic data of the candidate parking lots on the aggregation server, and decrypting the parking lot characteristic data to obtain a parking lot characteristic image matrix.
In one embodiment, the establishing an updated iterative model of a prediction score according to the user feature representation matrix, the score matrix and the parking lot feature representation matrix, iterating the updated iterative model to obtain a model gradient result, and sending the model gradient result to an aggregation server includes:
obtaining a prediction scoring parameter of each target user to the candidate parking lot according to the user characteristic image matrix of the target user and the parking lot characteristic image matrix, and obtaining an update iteration model for performing update iteration on the prediction scoring parameter according to a regularization least square method;
updating the feature vector and the model gradient result of the user feature image matrix in the updated iterative model, and sending the model gradient result in the updated iterative model to the aggregation server to instruct the aggregation server to update and calculate the parking lot feature image matrix according to the model gradient result to obtain the updated parking lot feature image matrix.
In one embodiment, the obtaining, according to the user characteristic image matrix of the target user and the parking lot characteristic image matrix, a prediction score parameter of each target user for the candidate parking lot, and according to a regularized least square method, obtaining an update iterative model for performing update iteration on the prediction score parameter includes:
acquiring a user characteristic portrait matrix of the target user, wherein the user characteristic portrait matrix contains user hidden attribute information;
performing inner product operation according to the user characteristic image matrix and the parking lot characteristic image matrix to obtain a prediction scoring parameter of the target user to the candidate parking lot;
and creating an updating iterative model for updating and iterating the prediction scoring parameters according to the user characteristic image matrix, the parking lot characteristic image matrix, the prediction scoring parameters and a regularized least square method.
In one embodiment, the iteratively updating the updated iterative model according to the updated parking lot feature image matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value to obtain a final prediction score value of the candidate parking lot includes:
inputting the updated parking lot feature portrait matrix into the updated iterative model for iterative updating;
and comparing the model gradient value of the updated iterative model obtained after each iterative update with a preset convergence threshold value, if the model gradient value is smaller than the preset convergence threshold value, determining that the model result of the updated iterative model reaches the minimum value, and taking the corresponding prediction scoring parameter value under the minimum value as the final prediction scoring value of the candidate parking lot.
In one embodiment, the scoring each candidate parking lot model according to the final predicted score value to obtain a score result, and recommending parking lots to the target user according to the score result includes:
performing weighted fusion calculation according to the final prediction score value and the current characteristic index data of each target user to obtain a score result of each candidate parking lot;
and arranging the candidate parking lots according to the grading result to obtain a grading sorting result, and selecting a target parking lot from the candidate parking lots according to the grading sorting result to serve as a recommended parking lot.
A parking lot recommendation device, the device comprising:
the acquisition module is used for acquiring a score matrix of the candidate parking lot and a parking lot characteristic image matrix; the scoring matrix comprises historical scores of each candidate parking lot by a target user;
the first updating iteration module is used for establishing an updating iteration model for predicting scores according to the user characteristic portrait matrix, the score matrix and the parking lot characteristic portrait matrix, iterating the updating iteration model to obtain a model gradient result, and sending the model gradient result to the aggregation server;
the receiving module is used for receiving the updated parking lot feature image matrix sent by the aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to the model gradient result;
the second updating iteration module is used for iteratively updating the updating iteration model according to the updated parking lot feature image matrix until the model gradient value of the updating iteration model is smaller than a preset convergence threshold value, and obtaining the final prediction score value of the candidate parking lot;
and the recommendation module is used for scoring each candidate parking lot model according to the final prediction scoring value to obtain a scoring result, and recommending parking lots to the target user according to the scoring result.
A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
acquiring a scoring matrix and a parking lot characteristic image matrix of a candidate parking lot; the scoring matrix comprises historical scores of each candidate parking lot by a target user;
establishing an updated iterative model of predictive score according to the user characteristic image matrix, the score matrix and the parking lot characteristic image matrix, iterating the updated iterative model to obtain a model gradient result, and sending the model gradient result to an aggregation server;
receiving the updated parking lot characteristic image matrix sent by the aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to the model gradient result;
iteratively updating the updated iterative model according to the updated parking lot feature image matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value, and obtaining a final prediction score value of the candidate parking lot;
and scoring each candidate parking lot model according to the final prediction scoring value to obtain a scoring result, and recommending parking lots to the target user according to the scoring result.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
acquiring a scoring matrix and a parking lot characteristic image matrix of the candidate parking lot; the scoring matrix comprises historical scores of each candidate parking lot by a target user;
establishing an updating iterative model for predicting scores according to the user characteristic portrait matrix, the scoring matrix and the parking lot characteristic portrait matrix, iterating the updating iterative model to obtain a model gradient result, and sending the model gradient result to an aggregation server;
receiving the updated parking lot characteristic image matrix sent by the aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to the model gradient result;
iteratively updating the updated iterative model according to the updated parking lot feature image matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value, and obtaining a final prediction score value of the candidate parking lot;
and scoring each candidate parking lot model according to the final prediction scoring value to obtain a scoring result, and recommending parking lots to the target user according to the scoring result.
The parking lot recommendation method, the parking lot recommendation device, the computer equipment and the storage medium acquire a score matrix and a parking lot characteristic image matrix of the candidate parking lot; the scoring matrix comprises historical scores of each candidate parking lot by a target user; establishing an updated iterative model of predictive score according to the user characteristic image matrix, the score matrix and the parking lot characteristic image matrix, iterating the updated iterative model to obtain a model gradient result, and sending the model gradient result to an aggregation server; receiving the updated parking lot feature image matrix sent by the aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to the model gradient result; iteratively updating the updated iterative model according to the updated parking lot feature image matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value, and obtaining a final prediction score value of the candidate parking lot; and scoring each candidate parking lot model according to the final prediction scoring value to obtain a scoring result, and recommending parking lots to the target user according to the scoring result. By adopting the method, accurate recommendation of the parking lot is realized under the condition of ensuring the personal privacy data safety of the user.
Drawings
FIG. 1 is a diagram of an exemplary environment in which a parking lot recommendation method may be implemented;
FIG. 2 is a schematic flow chart diagram of a parking lot recommendation method in one embodiment;
FIG. 3 is a flowchart illustrating the step of obtaining a scoring matrix according to one embodiment;
FIG. 4 is a diagram illustrating an exemplary process for normalizing a score matrix in an embodiment;
FIG. 5 is a flowchart illustrating steps for updating an iterative model to obtain a model gradient result in one embodiment;
FIG. 6 is a flowchart illustrating the steps of creating an updated iterative model in one embodiment;
FIG. 7 is a flowchart illustrating the steps of determining a final prediction score value in one embodiment;
FIG. 8 is a flowchart of the steps for deriving a recommended parking lot based on a final prediction score in one embodiment;
fig. 9 is a flowchart of a specific example of a parking lot recommendation method in one embodiment;
FIG. 10 is a block diagram showing the construction of a parking lot recommendation apparatus according to an embodiment;
FIG. 11 is a diagram illustrating an internal structure of a computer device in one embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of and not restrictive on the broad application.
The parking lot recommendation method provided by the application can be applied to the application environment shown in fig. 1. Wherein a terminal 102 (i.e., a target user-side computer device) communicates with an aggregation server 104 over a network. The terminal 102 acquires a scoring matrix and a parking lot characteristic image matrix of the candidate parking lot; the scoring matrix comprises historical scores of each candidate parking lot by a target user; then, the terminal 102 establishes an updated iterative model of predictive score according to the user characteristic image matrix, the score matrix and the parking lot characteristic image matrix, iterates the updated iterative model to obtain a model gradient result, and sends the model gradient result to the aggregation server; receiving the updated parking lot feature image matrix sent by the aggregation server 104; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to the model gradient result; the terminal 102 conducts iterative updating on the updated iterative model according to the updated parking lot feature image matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value, and a final prediction score value of the candidate parking lot is obtained; and finally, the terminal 102 scores each candidate parking lot model according to the final prediction scoring value to obtain a scoring result, and recommends the parking lot to the target user according to the scoring result. The terminal 102 may be, but is not limited to, various personal computers, notebook computers, smart phones, and tablet computers, and the aggregation server 104 may be implemented by an independent server or a server cluster formed by a plurality of servers.
In one embodiment, as shown in fig. 2, a parking lot recommendation method is provided, which is described by taking the method as an example applied to the terminal 102 in fig. 1, and includes the following steps:
step 201, acquiring a score matrix and a parking lot characteristic image matrix of a candidate parking lot; the scoring matrix comprises historical scores of each candidate parking lot of the target user.
Specifically, the scoring matrix (denoted by M) of the candidate parking lots includes the historical scores of the target user for each candidate parking lot, for example, the number of candidate parking lots is M, i.e., [ M ] = {1,2,3,... M }, and the historical scores of the target user a for the M candidate parking lots constitute a scoring matrix with 1 row and M columns. Meanwhile, the number of target users is also not limited to 1, and if there are n target users (i.e., [ n ] = {1,2,3.. N }), the dimension of the corresponding scoring matrix is n rows and M columns (i.e., M ∈ [ n ] × [ M ]). Each target user serves as a participant recommended by the parking lot and corresponds to one computer device, and a distributed system can be formed by the computer devices of the user terminals and the aggregation server.
In one implementation, the computer device corresponding to any target user acquires the scoring matrix and the parking lot feature image matrix (denoted by P) of the corresponding candidate parking lot. And the parking lot characteristic image matrix correspondingly comprises the characteristic data of the candidate parking lots. The parking lot feature data reflects hidden attribute information of a parking lot, for example, a d-dimensional feature vector of a j-th row in the parking lot feature imaging matrix represents a d-dimensional attribute feature hidden by a parking lot j.
Specifically, the computer device obtains a scoring matrix of a candidate parking lot corresponding to a target user of the computer device from the local storage space (or according to user verification input), then the computer device forwards the scoring matrix to the aggregation server according to a downloading request input by the user, and the encrypted parking lot feature data about the candidate parking lot is downloaded on the aggregation server.
Step 202, establishing an updating iterative model for predicting scores according to the user characteristic image matrix, the score matrix and the parking lot characteristic image matrix, iterating the updating iterative model to obtain a model gradient result, and sending the model gradient result to the aggregation server.
The user characteristic image matrix includes hidden attribute information of a target user, where the hidden attribute information is not a single data index, but may reflect information of behavior preference of the user, such as a behavior dependency relationship of the user (i.e., an event B is customized after an event a is done), a behavior time dependency relationship (an event is customized at a certain time), a user behavior transformation relationship, and the like, and this embodiment is not limited. Specifically, a d-dimensional user feature vector of the ith row in the user feature representation matrix U represents a d-dimensional attribute feature implied by the user i.
In one implementation, the computer device establishes an updated iterative model of predictive scoring according to the user characteristic portrait matrix, the scoring matrix and the parking lot characteristic portrait matrix, iterates the updated iterative model, obtains a model gradient result from each data in the iterative process, encrypts the model gradient result by using a homomorphic encryption method (for example, paillier), and sends the encrypted model gradient result to the aggregation server.
Specifically, the prediction score of the parking lot is the eigenvalue and eigenvector (i.e. the eigenvalue and eigenvector of the product of the user characteristic image matrix and the parking lot characteristic image matrix<u i ,p j >) Therefore, the update iteration of the prediction score may be performed by performing update iteration on the user characteristic image matrix and the parking lot characteristic image matrix, respectively.
Step 203, receiving the updated parking lot feature image matrix sent by the aggregation server; and the updated parking lot feature portrait matrix is obtained by performing iterative updating according to the model gradient result.
The updated parking lot feature portrait matrix is obtained by the aggregation server performing iterative updating on the received model gradient result and performing iteration according to the updated model gradient result. And after the aggregation server processes the updated parking lot feature image matrix, sending the parking lot feature image matrix to each target user (namely each recommended parking lot participant).
In an implementation, the (local) computer device receives the updated parking lot feature portrayal matrix sent by the aggregation server.
And 204, iteratively updating the updated iterative model according to the updated parking lot feature image matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value, and obtaining the final prediction score value of the candidate parking lot.
In implementation, the computer device continuously performs iterative updating on the updated iterative model according to the updated parking lot feature image matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold (represented by xi), the updated iterative model is determined to be converged, the model reaches a linear programming minimum value at the moment, and a corresponding prediction score value under the condition of the minimum value of the model is selected as a final prediction score value of the candidate parking lot.
And step 205, scoring each candidate parking lot model according to the final prediction scoring value to obtain a scoring result, and recommending parking lots to the target user according to the scoring result.
In implementation, the computer device scores each candidate parking lot model according to the final prediction scoring value to obtain a scoring result of each candidate parking lot, the scoring results can be ranked from large to small, and finally, a target parking lot meeting requirements is screened out from the candidate parking lots according to the ranking results and recommended to a target user.
In the parking lot recommendation method, a score matrix and a parking lot characteristic image matrix of a candidate parking lot are obtained; the scoring matrix comprises historical scores of each candidate parking lot for the target user; establishing an updating iterative model of predictive scores according to the user characteristic image matrix, the score matrix and the parking lot characteristic image matrix, iterating the updating iterative model to obtain a model gradient result, and sending the model gradient result to the aggregation server; receiving the updated parking lot characteristic image matrix sent by the aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to a model gradient result; iteratively updating the updated iterative model according to the updated parking lot feature portrait matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value, and obtaining a final prediction score value of the candidate parking lot; and grading each candidate parking lot model according to the final prediction grading value to obtain a grading result, and recommending parking lots to the target user according to the grading result. By adopting the method, accurate recommendation of the parking lot is realized under the condition of ensuring the personal privacy data safety of the user.
In one embodiment, as shown in fig. 3, the specific processing procedure of step 201 is as follows:
step 301, historical parking lot data of target users are obtained, and parking times data records of each target user for each candidate parking lot are obtained in the historical parking lot data in a statistical mode.
In implementation, the computer device acquires historical parking lot data of target users, and calculates parking number data records of each target user for each candidate parking lot in the historical parking lot data.
Specifically, if there are a plurality of target users participating in the parking lot recommendation, historical parking lot data (using D) of a plurality of target users (for example, n) are acquired ij And represents), statistics is performed on historical parking lot data of each target user, and each piece of historical parking lot data includes parking information of the user, for example, parking time of the user, unique identification (such as name of parking lot), position information (such as longitude and latitude information) of the parking lot, parking fee information, and the like (the number and type of data indexes included in the historical parking lot data are not limited in this embodiment). And then, counting the number of historical parking lot data with the same parking lot name and/or the same parking lot position information to obtain a parking number data record of each target user for each candidate parking lot.
Alternatively, if the target user does not pass through a certain candidate parking lot, the number of parking times for the candidate parking lot is recorded as 0.
Step 302, aiming at the parking times data record of each candidate parking lot of each target user, normalization processing is carried out to obtain the score value of each target user to each candidate parking lot.
In implementation, the computer device parks the vehicle for each candidate parking lot for each target userAnd recording the frequency data, and performing normalization processing to obtain the score value of each target user to each candidate parking lot. Specifically, the normalized number value is converted into a score value, which can be represented by r ij And expressing the concrete meaning of the score value of the ith target user on the jth candidate parking lot.
The specific process of digitizing the scoring matrix is shown in fig. 4, where Park represents parking lot side data, user represents target User side data, and the data of the parking times of each candidate parking lot corresponding to each target User is recorded and normalized (Normalization) is performed to obtain the processed scoring matrix.
And step 303, combining the score values of each target user to obtain a score matrix of each candidate parking lot.
In the implementation, the scoring values of each candidate parking lot of each target user are combined to obtain a scoring matrix of each candidate parking lot.
Specifically, for a local computer device of a target user, when the target user corresponds to the score values of m candidate parking lots, the score matrix is 1 row and m columns. If all parking lot participants contain n target users and M candidate parking lots, the overall scoring matrix M belongs to [ n ]]×[m]. The scoring matrix includes an element r ij Is shown as r ij ∈M。
Optionally, since the target user does not necessarily have a parking record in each candidate parking lot, the parking times and the corresponding score values of the candidate parking lots that the target user has not left are both 0, so that the overall score matrix of each parking lot participant is a sparse matrix, that is, a matrix with more elements than 0 in the matrix is not 0.
And step 304, downloading the encrypted parking lot characteristic data of the candidate parking lots on the aggregation server, and decrypting the parking lot characteristic data to obtain a parking lot characteristic image matrix.
In implementation, the computer device downloads the encrypted parking lot feature data of the candidate parking lots on the aggregation server, and then decrypts the parking lot feature data through a homomorphic encryption algorithm to obtain a decrypted parking lot feature portrait matrix P on the parking lot side.
In the embodiment, each candidate parking lot is scored according to the historical parking lot data of the target user, the corresponding scoring matrix is obtained, then each candidate parking lot can be predicted and scored according to the scoring matrix, accuracy of parking lot recommendation is improved, meanwhile, the target user does not need to send out privacy data, data processing is carried out locally, the encrypted parking lot data are downloaded on the parking lot side, and safety of the privacy data of the user is guaranteed.
In one embodiment, as shown in FIG. 5, the specific process of step 202 is as follows:
step 501, obtaining a prediction scoring parameter of each target user for a candidate parking lot according to the user characteristic image matrix of the target user and the parking lot characteristic image matrix, and obtaining an update iteration model for performing update iteration on the prediction scoring parameter according to a regularized least square method.
In implementation, the computer device obtains the prediction scoring parameter of each target user for the candidate parking lot according to the user characteristic image matrix U and the parking lot characteristic image matrix P of the target user<u i ,p j >And obtaining an updating iteration model for updating and iterating the prediction scoring parameters according to a regularized least square method. Specifically, the computer device needs to predict the score value (non-historical score value) of each target user to each candidate parking lot, and the matrix decomposition technology is to realize score prediction based on a bilinear model fitted on a score matrix (historical score), particularly to a prediction score parameter obtained by multiplying a user characteristic image matrix and a parking lot characteristic image matrix<u i ,p j >Therefore, this problem can be solved using a regularized least squares approach.
And 502, updating the feature vector and the model gradient result of the user feature image matrix in the updated iterative model, and sending the model gradient result in the updated iterative model to the aggregation server to instruct the aggregation server to update and calculate the parking lot feature image matrix according to the model gradient result to obtain the updated parking lot feature image matrix.
In implementation, when the local computer equipment of the target user updates the updated iterative model, the user feature portrait matrix in the updated iterative model is updated, and the t-th iteration feature vector of the user feature portrait matrix is updated
Figure BDA0002976967750000113
And model Gradient results (using Gradient) i That is, in the above-mentioned publication), wherein,
Figure BDA0002976967750000111
(the parameter specific meaning is see step 503 below), then the computer device will update the model Gradient result (Gradient) in the iterative model i ) And sending the parking lot feature image matrix to an aggregation server to instruct the aggregation server to update and calculate the parking lot feature image matrix according to the model gradient result so as to obtain an updated parking lot feature image matrix.
Specifically, the aggregation server receives the encrypted model Gradient result Gradient of each participant i And then, aggregating each model gradient result by using a homomorphic encryption security aggregation mode to ensure the privacy of data. Then, the aggregation server utilizes the aggregated model Gradient result Gradient i Updating the parking lot characteristic image matrix P, wherein the specific updating process is to perform iterative updating aiming at each characteristic vector in the parking lot characteristic image matrix, and the specific updating formula is as follows:
Figure BDA0002976967750000112
an update value representing the t-th update iteration for the j-th parking lot.
In the embodiment, the user participants (namely, each target user) encrypt the model gradient result of the model and send the encrypted model gradient result to the aggregation server, and the parking lot feature portrait matrix is updated and iterated through the aggregation server, so that the security of user privacy data is ensured.
In one embodiment, as shown in fig. 6, the specific processing procedure of step 501 is as follows:
step 601, obtaining a user characteristic portrait matrix of a target user, wherein the user characteristic portrait matrix contains user hidden attribute information.
In an implementation, the computer device obtains a user feature representation matrix of a target user, where the user feature representation matrix includes user hidden attribute information (also referred to as hidden attribute information), and the specific hidden attribute information is not limited in this embodiment.
And step 602, performing inner product operation according to the user characteristic image matrix and the parking lot characteristic image matrix to obtain a prediction scoring parameter of the target user to the candidate parking lot.
In implementation, the computer device performs inner product operation according to the user characteristic image matrix and the parking lot characteristic image matrix to obtain a prediction scoring parameter of the target user to the post-selected parking lot, namely the characteristic values of the user characteristic image matrix and the parking lot characteristic image matrix, and the prediction scoring parameter is used for calculating the feature value of the target user to the post-selected parking lot<u i ,p j >And (4) showing.
Step 603, creating an updating iterative model for updating and iterating the prediction scoring parameters according to the user characteristic representation matrix, the parking lot characteristic representation matrix, the prediction scoring parameters and the regularized least square method.
In implementation, the computer device creates an update iteration model for performing update iteration on the prediction scoring parameters according to the user characteristic representation matrix, the parking lot characteristic representation matrix, the prediction scoring parameters and the regularized least square method,
the expression for specifically updating the iterative model is as follows:
Figure BDA0002976967750000121
λ and μ are extremely small regularization factors for penalty terms, λ is a regularization factor of the user feature portrait matrix, and penalty terms are carried out on the matrix U; mu is a regularization factor of the parking lot feature image matrix, a penalty term is carried out on the matrix P, and the initial value of the penalty term is 1e-4, so that overfitting of the iterative model is prevented from being updated, and the generalization capability of the model is improved.
Figure BDA0002976967750000122
Representing the square of the 2 norm, is used to update U and P, i.e., their corresponding eigenvalues, using a stochastic gradient descent algorithm<u i ,p j >So that the updated iterative model (which may also be referred to as a least squares expression) converges when it reaches a minimum value. The updating process expression aiming at the user characteristic image matrix U and the parking lot characteristic image matrix P is as follows:
Figure BDA0002976967750000123
Figure BDA0002976967750000131
wherein
Figure BDA0002976967750000132
Figure BDA0002976967750000133
In particular, the amount of the solvent to be used,
Figure BDA0002976967750000134
the eigenvector value of the ith target user eigenvector of the user characteristic imaging matrix U at the t iteration,
Figure BDA0002976967750000135
and the feature vector value of the jth candidate parking lot feature vector of the parking lot feature image matrix in the t iteration is obtained.
Figure BDA0002976967750000136
And (4) gradient values of the ith target user feature vector of the user feature image matrix U in the current iteration gradient descent learning.
Figure BDA0002976967750000137
And (5) the j-th parking lot feature vector of the parking lot feature image matrix P is subjected to the t-th iteration.
In the embodiment, the updated iterative model is obtained by the least square method, so that the fitted final prediction score value is obtained under the condition that the model reaches the minimum value in the updating iterative process of the model, and the target user can predict the scores of all candidate parking lots.
In one embodiment, as shown in fig. 7, the specific processing procedure of step 204 is as follows:
and 701, inputting the updated parking lot feature portrait matrix into an updated iterative model for iterative updating.
In implementation, the computer device inputs the updated parking lot feature portrait matrix into the update iteration model for update iteration.
Step 702, comparing the model gradient value of the updated iterative model obtained after each iterative update with a preset convergence threshold value, if the model gradient value is smaller than the preset convergence threshold value, determining that the model result of the updated iterative model reaches the minimum value, and taking the corresponding prediction scoring parameter value under the minimum value as the final prediction scoring value of the candidate parking lot.
In implementation, the computer device compares the model gradient value (namely) of the updated iterative model obtained after each iterative update with a preset convergence threshold value, if the model gradient value is smaller than the preset convergence threshold value xi, the model convergence is proved, the model result of the updated iterative model is further determined to reach the minimum value, and the computer device takes the prediction scoring parameter value corresponding to the minimum value as the final prediction scoring value of the candidate parking lot. The specific update iteration process for updating the iterative model loops the above steps 202 to 203, and/or steps 501 to 502, and/or step 603, which is not limited in this embodiment.
In one embodiment, as shown in fig. 8, the specific processing procedure of step 205 is as follows:
and step 801, performing weighted fusion calculation according to the final prediction score value and the current characteristic index data of each target user to obtain a score result of each candidate parking lot.
In implementation, the computer device performs weighted fusion calculation according to the final prediction scoring value and the current characteristic index data of each target user to obtain a scoring result of each candidate parking lot.
For example, a score result (which may also be referred to as a composite score result) for each candidate parking lot is obtained by performing weighted average calculation on current feature indexes such as destination information, travel time, preference, and service to be provided of the user and the obtained final predicted score value according to preset weight indexes.
And step 802, arranging the candidate parking lots according to the grading result to obtain a grading sorting result, and selecting a target parking lot from the candidate parking lots as a recommended parking lot according to the grading sorting result.
In implementation, the participants are recommended to each parking lot, that is, the local computer device corresponding to each target user sorts the candidate parking lots according to the scoring result to obtain scoring sorting results of the candidate parking lots, and the target parking lots are screened out from all the candidate parking lots according to the scoring sorting results to serve as recommended parking lots, for example, the top 5 parking lots of the scoring sorting results are screened out to serve as final recommendation results, and the recommendation sequence of the recommendation results may be the same as the sequence corresponding to the scoring sorting results.
In the embodiment, by means of final prediction scoring, the current feature preference of the user is solved, each candidate parking lot is subjected to comprehensive scoring, and the target parking lot is determined to be the recommended parking lot according to the comprehensive scoring, so that the target parking lot not only contains all hidden attribute feature information contained in the target user historical record, but also refers to the current feature preference of the target user, and the accuracy of parking lot recommendation is improved.
It should be understood that although the various steps in the flowcharts of fig. 2-3, 5-8 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. 2-3, 5-8 may include multiple steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of execution of the steps or stages is not necessarily sequential, but may be alternated or performed with other steps or at least a portion of the steps or stages within other steps.
In an embodiment, as shown in fig. 9, an example of a parking lot recommendation method is provided, where a specific process is divided into three major parts, a technology of performing numerical scoring normalization on a parking lot historical data record of a target user to obtain a scoring matrix of the target user for a candidate parking lot, then, for each parking lot user participant, a distributed encryption matrix decomposition federal learning technology (also referred to as a parking lot federal technology based on an encrypted federal learning neural network model) is used, that is, an update iteration model is created by using a regularized least square method, a final prediction score value is determined by a mode of interacting a model gradient result with an aggregation server (the interaction process uses homomorphic encryption to achieve safe aggregation of the aggregation server), and finally, the target parking lot is recommended based on the distributed encryption matrix decomposition prediction (that is, the final prediction score value).
In one embodiment, as shown in fig. 10, there is provided a parking lot recommendation device 1000 including: an obtaining module 1010, a first update iteration module 1020, a receiving module 1030, a second update iteration module 1040, and a recommending module 1050, wherein:
the obtaining module 1010 is configured to obtain a score matrix of the candidate parking lot and a parking lot feature image matrix; the scoring matrix comprises historical scores of each candidate parking lot for the target user;
the first updating iteration module 1020 is used for establishing an updating iteration model for predicting scores according to the user characteristic sketch matrix, the score matrix and the parking lot characteristic sketch matrix, iterating the updating iteration model to obtain a model gradient result, and sending the model gradient result to the aggregation server;
a receiving module 1030, configured to receive the updated parking lot feature image matrix sent by the aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to a model gradient result;
the second updating iteration module 1040 is configured to iteratively update the updated iteration model according to the updated parking lot feature representation matrix until the model gradient value of the updated iteration model is smaller than a preset convergence threshold value, so as to obtain a final prediction score value of the candidate parking lot;
and the recommendation module 1050 is configured to score each candidate parking lot model according to the final prediction score value to obtain a score result, and recommend a parking lot to the target user according to the score result.
In one embodiment, the obtaining module 1010 is specifically configured to obtain historical parking lot data of target users, and obtain a parking number data record of each target user for each candidate parking lot in the historical parking lot data by statistics;
carrying out normalization processing on the parking time data records of each candidate parking lot of each target user to obtain the score value of each target user on each candidate parking lot;
combining the score values of each target user to obtain a score matrix of each candidate parking lot;
and downloading the encrypted parking lot characteristic data of the candidate parking lot on the aggregation server, and decrypting the parking lot characteristic data to obtain a parking lot characteristic image matrix.
In one embodiment, the first update iteration module 1020 is specifically configured to obtain a prediction scoring parameter of each target user for a candidate parking lot according to the user feature image matrix of the target user and the parking lot feature image matrix, and obtain an update iteration model for performing update iteration on the prediction scoring parameter according to a regularized least square method;
and updating the feature vector and the model gradient result of the user feature image matrix in the updated iterative model, and sending the model gradient result in the updated iterative model to the aggregation server so as to instruct the aggregation server to update and calculate the parking lot feature image matrix according to the model gradient result, thereby obtaining the updated parking lot feature image matrix.
In one embodiment, the first update iteration module 1020 is further configured to obtain a user feature representation matrix of the target user, where the user feature representation matrix includes user hidden attribute information;
performing inner product operation according to the user characteristic image matrix and the parking lot characteristic image matrix to obtain a prediction scoring parameter of the target user to the candidate parking lot;
and creating an updating iterative model for updating and iterating the prediction scoring parameters according to the user characteristic representation matrix, the parking lot characteristic representation matrix, the prediction scoring parameters and the regularized least square method.
In an embodiment, the second update iteration module 1040 is specifically configured to input the updated parking lot feature representation matrix into the update iteration model for iterative update;
and comparing the model gradient value of the updated iterative model obtained after each iterative update with a preset convergence threshold value, if the model gradient value is smaller than the preset convergence threshold value, determining that the model result of the updated iterative model reaches the minimum value, and taking the corresponding prediction scoring parameter value under the minimum value as the final prediction scoring value of the candidate parking lot.
In one embodiment, the recommending module 1050 is specifically configured to perform weighted fusion calculation according to the final predicted score value and the current feature index data of each target user to obtain a score result of each candidate parking lot;
and arranging the candidate parking lots according to the grading result to obtain a grading sorting result, and selecting a target parking lot from the candidate parking lots as a recommended parking lot according to the grading sorting result.
For specific definition of the parking lot recommendation device, reference may be made to the above definition of the parking lot recommendation method, which is not described herein again. The modules in the parking lot recommendation device can be wholly or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 11. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used for communicating with an external terminal in a wired or wireless manner, and the wireless manner can be realized through WIFI, an operator network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement a parking lot recommendation method. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on a shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.
Those skilled in the art will appreciate that the architecture shown in fig. 11 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
acquiring a scoring matrix and a parking lot characteristic image matrix of a candidate parking lot; the scoring matrix comprises historical scores of each candidate parking lot by the target user;
establishing an updating iterative model of predictive scores according to the user characteristic image matrix, the score matrix and the parking lot characteristic image matrix, iterating the updating iterative model to obtain a model gradient result, and sending the model gradient result to the aggregation server;
receiving an updated parking lot characteristic image matrix sent by an aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to a model gradient result;
iteratively updating the updated iterative model according to the updated parking lot feature portrait matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value, and obtaining a final prediction score value of the candidate parking lot;
and grading each candidate parking lot model according to the final prediction grading value to obtain a grading result, and recommending parking lots to the target user according to the grading result.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
acquiring historical parking lot data of target users, and statistically acquiring parking time data records of each target user for each candidate parking lot in the historical parking lot data;
carrying out normalization processing on the parking time data records of each candidate parking lot of each target user to obtain the score value of each target user on each candidate parking lot;
combining the score values of each target user to obtain a score matrix of each candidate parking lot;
and downloading the encrypted parking lot characteristic data of the candidate parking lots on the aggregation server, and decrypting the parking lot characteristic data to obtain a parking lot characteristic image matrix.
In one embodiment, the processor when executing the computer program further performs the steps of:
obtaining a prediction scoring parameter of each target user to a candidate parking lot according to the user characteristic image matrix of the target user and the parking lot characteristic image matrix, and obtaining an update iteration model for updating and iterating the prediction scoring parameter according to a regularized least square method;
and updating the feature vector and the model gradient result of the user feature image matrix in the updated iterative model, and sending the model gradient result in the updated iterative model to the aggregation server so as to instruct the aggregation server to update and calculate the parking lot feature image matrix according to the model gradient result to obtain the updated parking lot feature image matrix.
In one embodiment, the processor when executing the computer program further performs the steps of:
acquiring a user characteristic portrait matrix of a target user, wherein the user characteristic portrait matrix contains user hidden attribute information;
performing inner product operation according to the user characteristic image matrix and the parking lot characteristic image matrix to obtain a prediction scoring parameter of the target user to the candidate parking lot;
and creating an updating iterative model for updating and iterating the prediction scoring parameters according to the user characteristic representation matrix, the parking lot characteristic representation matrix, the prediction scoring parameters and the regularized least square method.
In one embodiment, the processor when executing the computer program further performs the steps of:
inputting the updated parking lot feature portrait matrix into an update iteration model for iteration update;
and comparing the model gradient value of the updated iterative model obtained after each iterative update with a preset convergence threshold value, if the model gradient value is smaller than the preset convergence threshold value, determining that the model result of the updated iterative model reaches the minimum value, and taking the corresponding prediction scoring parameter value under the minimum value as the final prediction scoring value of the candidate parking lot.
In one embodiment, the processor when executing the computer program further performs the steps of:
performing weighted fusion calculation according to the final prediction scoring value and the current characteristic index data of each target user to obtain a scoring result of each candidate parking lot;
and arranging the candidate parking lots according to the grading result to obtain a grading sorting result, and selecting a target parking lot from the candidate parking lots as a recommended parking lot according to the grading sorting result.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring a scoring matrix and a parking lot characteristic image matrix of the candidate parking lot; the scoring matrix comprises historical scores of each candidate parking lot by the target user;
establishing an updating iterative model of predictive scores according to the user characteristic image matrix, the score matrix and the parking lot characteristic image matrix, iterating the updating iterative model to obtain a model gradient result, and sending the model gradient result to the aggregation server;
receiving the updated parking lot characteristic image matrix sent by the aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to a model gradient result;
iteratively updating the updated iterative model according to the updated parking lot feature portrait matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value, and obtaining a final prediction score value of the candidate parking lot;
and grading each candidate parking lot model according to the final prediction grading value to obtain a grading result, and recommending parking lots to the target user according to the grading result.
In one embodiment, the computer program when executed by the processor further performs the steps of:
acquiring historical parking lot data of target users, and statistically acquiring parking time data records of each target user for each candidate parking lot in the historical parking lot data;
carrying out normalization processing on the parking time data records of each candidate parking lot of each target user to obtain the score value of each target user on each candidate parking lot;
combining the score values of each target user to obtain a score matrix of each candidate parking lot;
and downloading the encrypted parking lot characteristic data of the candidate parking lots on the aggregation server, and decrypting the parking lot characteristic data to obtain a parking lot characteristic image matrix.
In one embodiment, the computer program when executed by the processor further performs the steps of:
obtaining a prediction scoring parameter of each target user to a candidate parking lot according to the user characteristic image matrix of the target user and the parking lot characteristic image matrix, and obtaining an update iteration model for updating and iterating the prediction scoring parameter according to a regularized least square method;
and updating the feature vector and the model gradient result of the user feature image matrix in the updated iterative model, and sending the model gradient result in the updated iterative model to the aggregation server so as to instruct the aggregation server to update and calculate the parking lot feature image matrix according to the model gradient result to obtain the updated parking lot feature image matrix.
In one embodiment, the computer program when executed by the processor further performs the steps of:
acquiring a user characteristic portrait matrix of a target user, wherein the user characteristic portrait matrix contains user hidden attribute information;
performing inner product operation according to the user characteristic image matrix and the parking lot characteristic image matrix to obtain a prediction scoring parameter of the target user to the candidate parking lot;
and creating an updating iterative model for updating and iterating the prediction scoring parameters according to the user characteristic image matrix, the parking lot characteristic image matrix, the prediction scoring parameters and the regularized least square method.
In one embodiment, the computer program when executed by the processor further performs the steps of:
inputting the updated parking lot feature portrait matrix into an update iteration model for iteration update;
and comparing the model gradient value of the updated iterative model obtained after each iterative update with a preset convergence threshold value, if the model gradient value is smaller than the preset convergence threshold value, determining that the model result of the updated iterative model reaches the minimum value, and taking the corresponding prediction scoring parameter value under the minimum value as the final prediction scoring value of the candidate parking lot.
In one embodiment, the computer program when executed by the processor further performs the steps of:
performing weighted fusion calculation according to the final predicted scoring value and the current characteristic index data of each target user to obtain a scoring result of each candidate parking lot;
and arranging the candidate parking lots according to the grading result to obtain a grading sorting result, and selecting a target parking lot from the candidate parking lots as a recommended parking lot according to the grading sorting result.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (10)

1. A parking lot recommendation method, the method comprising:
acquiring a scoring matrix and a parking lot characteristic image matrix of the candidate parking lot; the scoring matrix comprises historical scores of each candidate parking lot by a target user;
establishing an updating iterative model for predicting scores according to the user characteristic portrait matrix, the scoring matrix and the parking lot characteristic portrait matrix, iterating the updating iterative model to obtain a model gradient result, and sending the model gradient result to an aggregation server;
receiving the updated parking lot characteristic image matrix sent by the aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to the model gradient result;
iteratively updating the updated iterative model according to the updated parking lot feature image matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value, and obtaining a final prediction score value of the candidate parking lot;
and scoring each candidate parking lot model according to the final prediction scoring value to obtain a scoring result, and recommending parking lots to the target user according to the scoring result.
2. The method of claim 1, wherein the obtaining of the scoring matrix and the parking lot feature imaging matrix of the candidate parking lot comprises:
acquiring historical parking lot data of target users, and counting in the historical parking lot data to obtain parking number data records of each target user for each candidate parking lot;
carrying out normalization processing on the parking time data records of each candidate parking lot of each target user to obtain the score value of each candidate parking lot of each target user;
merging the scoring values of each target user to obtain a scoring matrix of each candidate parking lot;
and downloading the encrypted parking lot characteristic data of the candidate parking lots on the aggregation server, and decrypting the parking lot characteristic data to obtain a parking lot characteristic image matrix.
3. The method of claim 1, wherein the establishing of an updated iterative model of predictive scoring according to the user feature representation matrix, the scoring matrix and the parking lot feature representation matrix, the iterating of the updated iterative model to obtain a model gradient result, and the sending of the model gradient result to an aggregation server comprises:
obtaining a prediction scoring parameter of each target user to the candidate parking lot according to the user characteristic image matrix of the target user and the parking lot characteristic image matrix, and obtaining an update iteration model for performing update iteration on the prediction scoring parameter according to a regularization least square method;
updating the feature vector and the model gradient result of the user feature image matrix in the updated iterative model, and sending the model gradient result in the updated iterative model to the aggregation server to instruct the aggregation server to update and calculate the parking lot feature image matrix according to the model gradient result to obtain an updated parking lot feature image matrix.
4. The method according to claim 3, wherein the obtaining of the prediction scoring parameter of each target user for the candidate parking lot according to the user characteristic image matrix of the target user and the parking lot characteristic image matrix and obtaining of the update iterative model for performing update iteration on the prediction scoring parameter according to a regularized least square method comprises:
acquiring a user characteristic portrait matrix of the target user, wherein the user characteristic portrait matrix contains user hidden attribute information;
performing inner product operation according to the user characteristic image matrix and the parking lot characteristic image matrix to obtain a prediction scoring parameter of the target user for the candidate parking lot;
and creating an updating iterative model for updating and iterating the prediction scoring parameters according to the user characteristic image matrix, the parking lot characteristic image matrix, the prediction scoring parameters and a regularization least square method.
5. The method according to claim 1, wherein the iteratively updating the updated iterative model according to the updated parking lot feature image matrix until the model gradient value of the updated iterative model is smaller than a preset convergence threshold value to obtain a final prediction score value of the candidate parking lot, includes:
inputting the updated parking lot feature portrait matrix into the updated iterative model for iterative update;
and comparing the model gradient value of the updated iterative model obtained after each iterative update with a preset convergence threshold value, if the model gradient value is smaller than the preset convergence threshold value, determining that the model result of the updated iterative model reaches the minimum value, and taking the corresponding prediction scoring parameter value under the minimum value as the final prediction scoring value of the candidate parking lot.
6. The method according to claim 1, wherein the scoring each candidate parking lot model according to the final predicted score to obtain a scoring result, and performing parking lot recommendation to the target user according to the scoring result comprises:
performing weighted fusion calculation according to the final prediction score value and the current characteristic index data of each target user to obtain a score result of each candidate parking lot;
and arranging the candidate parking lots according to the grading result to obtain a grading sorting result, and selecting a target parking lot from the candidate parking lots according to the grading sorting result to serve as a recommended parking lot.
7. A parking lot recommendation device, the device comprising:
the acquisition module is used for acquiring a scoring matrix of the candidate parking lot and a parking lot characteristic image matrix; the scoring matrix comprises historical scores of each candidate parking lot by a target user;
the first updating iteration module is used for establishing an updating iteration model of the prediction score according to the user characteristic image matrix, the score matrix and the parking lot characteristic image matrix, iterating the updating iteration model to obtain a model gradient result, and sending the model gradient result to the aggregation server;
the receiving module is used for receiving the updated parking lot feature image matrix sent by the aggregation server; the updated parking lot feature portrait matrix is obtained by performing iterative updating according to the model gradient result;
the second updating iteration module is used for carrying out iteration updating on the updating iteration model according to the updated parking lot feature image matrix until the model gradient value of the updating iteration model is smaller than a preset convergence threshold value, and obtaining the final prediction score value of the candidate parking lot;
and the recommendation module is used for scoring each candidate parking lot model according to the final prediction scoring value to obtain a scoring result, and recommending parking lots to the target user according to the scoring result.
8. The device according to claim 7, wherein the first update iteration module is further configured to obtain a prediction scoring parameter of each target user for the candidate parking lot according to the user feature image matrix of the target user and the parking lot feature image matrix, and obtain an update iteration model for performing update iteration on the prediction scoring parameter according to a regularized least square method;
updating the feature vector and the model gradient result of the user feature image matrix in the updated iterative model, and sending the model gradient result in the updated iterative model to the aggregation server to instruct the aggregation server to update and calculate the parking lot feature image matrix according to the model gradient result to obtain an updated parking lot feature image matrix.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 6.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.
CN202110276770.9A 2021-03-15 2021-03-15 Parking lot recommendation method and device, computer equipment and storage medium Active CN113032670B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110276770.9A CN113032670B (en) 2021-03-15 2021-03-15 Parking lot recommendation method and device, computer equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110276770.9A CN113032670B (en) 2021-03-15 2021-03-15 Parking lot recommendation method and device, computer equipment and storage medium

Publications (2)

Publication Number Publication Date
CN113032670A CN113032670A (en) 2021-06-25
CN113032670B true CN113032670B (en) 2022-10-25

Family

ID=76468830

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110276770.9A Active CN113032670B (en) 2021-03-15 2021-03-15 Parking lot recommendation method and device, computer equipment and storage medium

Country Status (1)

Country Link
CN (1) CN113032670B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114926287A (en) * 2022-06-02 2022-08-19 河北源达信息技术股份有限公司 Method and device for intelligently recommending investment products and electronic equipment
CN117829377A (en) * 2024-03-04 2024-04-05 德阳城市智慧之心信息技术有限公司 Neural network-based remaining parking space prediction method, device, equipment and medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107563841A (en) * 2017-08-03 2018-01-09 电子科技大学 A kind of commending system decomposed that scored based on user
CN110110226A (en) * 2019-04-18 2019-08-09 国家电网有限公司 A kind of proposed algorithm, recommender system and terminal device
CN112380439A (en) * 2020-11-19 2021-02-19 平安科技（深圳）有限公司 Target object recommendation method and device, electronic equipment and computer-readable storage medium

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107563841A (en) * 2017-08-03 2018-01-09 电子科技大学 A kind of commending system decomposed that scored based on user
CN110110226A (en) * 2019-04-18 2019-08-09 国家电网有限公司 A kind of proposed algorithm, recommender system and terminal device
CN112380439A (en) * 2020-11-19 2021-02-19 平安科技（深圳）有限公司 Target object recommendation method and device, electronic equipment and computer-readable storage medium

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于用户分类的隐含因子模型研究;黎新志等;《计算机应用研究》;20170721(第08期);全文 *

Also Published As

Publication number Publication date
CN113032670A (en) 2021-06-25

Similar Documents

Publication Publication Date Title
Li et al. Rank-geofm: A ranking based geographical factorization method for point of interest recommendation
CN113032670B (en) Parking lot recommendation method and device, computer equipment and storage medium
CN112585620A (en) Distributed tagging for supervised learning
US20220067181A1 (en) Methods and systems for secure data analysis and machine learning
CN111625809B (en) Data authorization method and device, electronic equipment and storage medium
Ng et al. Machine learning for a London housing price prediction mobile application
CN111553744A (en) Federal product recommendation method, device, equipment and computer storage medium
CN109741035A (en) Data processing method, device, computer equipment and storage medium
US20180118164A1 (en) Customizable Vehicle Security System
CN115618008A (en) Account state model construction method and device, computer equipment and storage medium
Singh et al. Learning privacy habits of pds owners
US11409909B1 (en) Data privacy enforcers
US20210373937A1 (en) Multiparty computations
CN113158047A (en) Recommendation model training and information pushing method, device, equipment and medium
CN117349899A (en) Sensitive data processing method, system and storage medium based on forgetting model
CN116805039A (en) Feature screening method, device, computer equipment and data disturbance method
CN110990859B (en) Method and system for executing machine learning under data privacy protection
CN111402069A (en) Accumulation fund data acquisition method and device, computer equipment and storage medium
Wei et al. CrimeSTC: A deep spatial-temporal-categorical network for citywide crime prediction
CN113112312B (en) Method, apparatus and computer-readable storage medium for generating a model for a user
CN115758271A (en) Data processing method, data processing device, computer equipment and storage medium
CN115455276A (en) Method and device for recommending object, computer equipment and storage medium
Errounda Adaptive Differential Privacy for Decentralized Mobility Data Sharing and Forecasting
CN113468561B (en) Data protection method, device and server
Zhang Location and Data Privacy Preservation in Intelligent Systems

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