WO2020151725A1 - Procédé et dispositif de prédiction d'emplacement - Google Patents

Procédé et dispositif de prédiction d'emplacement Download PDF

Info

Publication number
WO2020151725A1
WO2020151725A1 PCT/CN2020/073652 CN2020073652W WO2020151725A1 WO 2020151725 A1 WO2020151725 A1 WO 2020151725A1 CN 2020073652 W CN2020073652 W CN 2020073652W WO 2020151725 A1 WO2020151725 A1 WO 2020151725A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
predicted
sample
feature value
historical behavior
Prior art date
Application number
PCT/CN2020/073652
Other languages
English (en)
Chinese (zh)
Inventor
谢君
卓呈祥
Original Assignee
北京嘀嘀无限科技发展有限公司
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 北京嘀嘀无限科技发展有限公司 filed Critical 北京嘀嘀无限科技发展有限公司
Publication of WO2020151725A1 publication Critical patent/WO2020151725A1/fr

Links

Images

Classifications

    • 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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • 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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G06Q30/0202—Market predictions or forecasting for commercial activities
    • 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
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • 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
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0607—Regulated
    • 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
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0633—Lists, e.g. purchase orders, compilation or processing
    • G06Q30/0635—Processing of requisition or of purchase orders
    • 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
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0639—Item locations

Definitions

  • This application relates to the field of machine learning technology, and in particular to a position prediction method and device.
  • the user's location location information can be obtained in real time, but the method of obtaining the location location in real time has poor timeliness, which is not conducive to early resource allocation and the configuration of related service strategies.
  • the purpose of the embodiments of the present application is to provide a location prediction method and device, which can determine the location of the user to be predicted within a preset time period in the future with higher accuracy.
  • a position prediction method may include: acquiring features associated with the historical behavior of the user to be predicted, the features including at least location information related to the historical behavior of the user to be predicted; at least based on the features associated with the historical behavior of the user to be predicted, through pre-trained
  • the location prediction model obtains the location prediction result of the user to be predicted in the future preset time period.
  • the feature associated with the historical behavior of the user to be predicted may include the first feature value of the user to be predicted under at least one historical behavior feature, and the user to be predicted within a preset time period in the future The second feature value under the target time feature and the third feature value under the target location feature.
  • the location prediction result of the user to be predicted in the future preset time period may include the probability that the user to be predicted will go to the area to be predicted in the future preset time period;
  • the first feature reflects the The distribution information of the at least one historical behavior in different regions, the distribution information of the at least one historical behavior in time, the association information between the at least one historical behavior and the future preset time period, the at least one historical behavior and One or a combination of one or more of the associated information of the region to be predicted;
  • the second feature reflects the attribute information of the future preset time period;
  • the third feature reflects the attribute information of the region to be predicted.
  • the acquiring the first feature value of the user to be predicted under at least one historical behavior feature includes: extracting the user to be predicted from the at least one historical behavior information of the user to be predicted Regional historical behavior information of the region, and extracting the regional time historical behavior information of the user to be predicted in the region to be predicted in each historical time period among the multiple historical time periods; according to the at least one user to be predicted
  • the first feature value under the first historical behavior feature of the user to be predicted is determined based on the historical behavior information, the regional historical behavior information, and the regional time historical behavior information; This kind of historical behavior information determines the first feature value under the second historical behavior feature of the user to be predicted.
  • the first historical behavior feature may include one or more of the following: the number of occurrences of historical behaviors corresponding to multiple POI classifications of points of interest, and the number of occurrences of historical behaviors per day within a preset time period. , The number of occurrences of historical behaviors on working days, the number of occurrences of historical behaviors on non-working days, and the number of occurrences corresponding to different historical behaviors.
  • the second historical behavior feature may include one or more of the following: whether the area where the last historical behavior occurred is the area to be predicted, and the time between the last historical behavior and the time to be predicted The time interval, the POI classification of the destination of the last historical behavior, the POI classification of the departure place of the last historical behavior, and the number of areas reached by the user to be predicted.
  • the historical behavior information may include one or more of historical bubbling behavior information, historical billing behavior information, and historical billing behavior information.
  • the method may further include: obtaining the user attribute characteristics of the user to be predicted; and obtaining the user attribute characteristics of the user to be predicted through a pre-trained location prediction model based at least on the characteristics associated with the historical behavior of the user to be predicted
  • the position prediction result of the user in the future preset time period includes: obtaining the to-be-predicted through a pre-trained position prediction model based on the characteristics associated with the historical behavior of the user to be predicted and the user attribute characteristics of the user to be predicted The location prediction result of the user in the future preset time period.
  • the user attribute characteristics may include one or more of the following: the number of orders, whether it is a business person, whether it is a tourist, the area where the user's home is located, and the area where the user company is located.
  • the following method may be used to train the position prediction model: obtaining the first sample feature value of a plurality of sample users under at least one historical behavior characteristic, and the historical preset time of each sample user In the segment, the second sample feature value under the target time feature and the third sample feature value under the target location feature; the sample users include positive sample users and negative sample users; based on the first sample of each sample user This feature value, the second sample feature value, and the third sample feature value are used to train the position prediction model.
  • the second sample feature value and the third sample feature value under the target location feature include: for each sample region to be predicted in the plurality of sample regions to be predicted, obtaining a sample user corresponding to the sample region to be predicted in at least one
  • the training of the location prediction model based on the first sample feature value, the second sample feature value, and the third sample feature value of each sample user includes: based on each sample area to be predicted
  • the first sample feature value, the second sample feature value, and the third sample feature value of the corresponding sample user are trained to train the position prediction model.
  • the training of the location prediction model based on the first sample feature value of each sample user and the second sample feature value includes: based on the first sample feature value of each sample user This feature value, the second sample feature value, and the third sample feature value construct multiple sub-decision trees; determine the multiple sub-decision trees as the position prediction model.
  • constructing multiple sub-decision trees based on the first sample feature value, the second sample feature value, and the third sample feature value of each sample user includes: Among the historical behavior characteristics, the target time characteristics, and the target area characteristics, multiple target characteristics are randomly determined; based on the characteristic values of the sample users under the target characteristics, a sub-decision tree of the current iteration cycle is constructed ; The sub-decision tree based on the current iteration cycle and the sub-decision tree of the historical iteration cycle constitute the current decision tree set, and determine the loss of the current decision tree set; if the loss is greater than the preset loss threshold, complete the current iteration Cycle, and return to the step of randomly determining multiple target characteristics among the historical behavior characteristics, the target time characteristics, the target area characteristics, and the user attribute characteristics; if the loss is not greater than the preset loss threshold In this case, the current decision tree set is determined as the position prediction model.
  • the determining the loss of the current decision tree set includes: comparing the first test sample feature value of a plurality of test sample users under the historical behavior characteristics, and the time to be predicted for each test sample user The feature and the feature value of the second test sample under the feature of the region to be predicted are input into the current decision tree set to obtain the location prediction result corresponding to each test sample user; based on the corresponding test sample user The location prediction result, and the corresponding actual location, determine the loss of the current decision tree set.
  • the method further includes: obtaining a fourth sample feature value of each of the sample users under multiple user attribute features;
  • the training of the position prediction model based on the first sample feature value, the second sample feature value, and the third sample feature value of each sample user includes: the The first sample feature value, the second sample feature value, the third sample feature value, and the fourth sample feature value are used to train the position prediction model.
  • the first feature value, the second feature value, and the third feature value are input into a pre-trained location prediction model to obtain the user to be predicted at the preset time in the future
  • the location prediction result of the segment includes: inputting the first feature value, the second feature value, and the third feature value into a pre-trained location prediction model to obtain the sub-prediction results of each sub-decision tree; The sub-prediction results of each sub-decision tree are weighted and summed to determine the location prediction result of the user to be predicted in the future time period.
  • the feature associated with the historical behavior of the user to be predicted reflects one or more historical ride-hailing information of the user to be predicted.
  • the historical ride-hailing information includes one or more of the following information: departure place, destination, departure place attributes, destination attributes, and time information.
  • it further includes acquiring the user attribute characteristics of the user to be predicted and the area to be predicted; the location prediction result of the user to be predicted in the future preset time period includes the user to be predicted at a preset time in the future The probability of going to the area to be predicted within a segment; said at least based on the features associated with the historical behavior of the user to be predicted, obtaining the location prediction result of the user to be predicted in the future preset time period through a pre-trained location prediction model , Including: based on the features associated with the historical behavior of the user to be predicted, the user attribute feature of the user to be predicted, the future preset time period, and the area to be predicted, obtaining the to-be-predicted location through a pre-trained location prediction model Predict the location prediction result of the user in the future preset time period.
  • the location prediction is based on the features associated with the historical behavior of the user to be predicted, the user attribute features of the user to be predicted, the future preset time period, and the area to be predicted.
  • the model obtains the location prediction result of the user to be predicted in the future preset time period, including, through the location prediction model: processing the user attribute characteristics of the user to be predicted, the future preset time period, and the The region to be predicted obtains the first vector representation; the user attribute characteristics are respectively fused with each historical taxi information to obtain the vector representation corresponding to each historical taxi information; the first vector representation and each time are processed based on the attention mechanism
  • the vector representation corresponding to the historical taxi information obtains the second vector representation; based on the second vector representation, the probability of the user to be predicted going to the area to be predicted in the future preset time period is determined.
  • a position prediction system includes: at least one storage medium including a set of instructions for position prediction; and at least one processor in communication with the storage medium, wherein when the set of instructions is executed, the at least one processor uses ⁇ : Obtaining features associated with the historical behavior of the user to be predicted; the features include at least location information related to the historical behavior of the user to be predicted; at least based on the features associated with the historical behavior of the user to be predicted, through a pre-trained location prediction model Obtain the location prediction result of the user to be predicted in the future preset time period.
  • a position prediction device includes: an acquisition module, configured to acquire features associated with the historical behavior of the user to be predicted; the features include at least location information related to the historical behavior of the user to be predicted; For the associated features, the position prediction result of the user to be predicted in the future preset time period is obtained through a pre-trained position prediction model.
  • a computer-readable storage medium characterized in that a computer program is stored on the computer-readable storage medium, and the computer program is executed when the processor is running: obtaining and predicting user history Behavior-related features; the features include at least location information related to the historical behavior of the user to be predicted; at least based on the features associated with the historical behavior of the user to be predicted, the location of the user to be predicted is acquired through a pre-trained location prediction model Describe the location prediction result for a preset time period in the future.
  • a position prediction method includes: obtaining a first feature value of a user to be predicted under at least one historical behavior feature, and a second feature value of the user to be predicted under a target time feature and a target location within a preset time period in the future.
  • the third feature value under the feature; the first feature value, the second feature value, and the third feature value are input into a pre-trained location prediction model to obtain the future prediction of the user to be predicted Set the position prediction result of the time period.
  • a location prediction device includes: an acquisition module for acquiring the first characteristic value of the user to be predicted under at least one historical behavior characteristic, and the second characteristic of the user to be predicted under the target time characteristic in a preset time period in the future Value and the third feature value under the target location feature; a prediction module for inputting the first feature value, the second feature value, and the third feature value into a pre-trained location prediction model to obtain The location prediction result of the user to be predicted in the future preset time period.
  • Fig. 1 is a schematic diagram of an exemplary online platform of an exemplary on-demand service system according to some embodiments of the present application
  • Figure 2A is a schematic diagram of exemplary hardware and/or software components of a computing device according to some embodiments of the present application
  • 2B is a schematic diagram of exemplary hardware and/or software components of a mobile device according to some embodiments of the present application.
  • FIG. 3 shows a flowchart of an exemplary process for predicting a position according to some embodiments of the present application
  • FIG. 4 shows a flowchart of a process of acquiring a first characteristic value under a first historical behavior characteristic and a first characteristic value under a second historical behavior characteristic according to some embodiments of the present application;
  • Fig. 5 shows a flowchart of a process of training a position prediction model according to some embodiments of the present application
  • Fig. 6 shows a flowchart of a process for training a position prediction model according to some embodiments of the present application
  • FIG. 7A shows a flowchart of an exemplary process for predicting a position according to some embodiments of the present application
  • Fig. 7B shows a flowchart of a prediction process of a position prediction model according to some embodiments of the present application.
  • Fig. 8 shows an exemplary block diagram of a position prediction apparatus according to some embodiments of the present application.
  • the means of transportation of the transportation system can include taxis, private cars, downwinds, buses, trains, bullet trains, high-speed railways, subways, ships, airplanes, spacecraft, hot air balloons, or unmanned vehicles, etc., or any of them combination.
  • This application may also include any service system for providing services to users based on the Internet, for example, a system for sending and/or receiving express delivery, and a service system for transactions between buyers and sellers.
  • the application of the system or method of the present application may include web pages, browser plug-ins, client terminals, customized systems, internal analysis systems, or artificial intelligence robots, etc., or any combination thereof.
  • This application relates to a location prediction method.
  • the method can input the features associated with the historical behavior of the user to be predicted into a pre-trained location prediction model, and predict whether the user to be predicted will appear at the target location in a preset time period in the future, so that the user location can be predicted in advance Information is conducive to the deployment of resources in advance and the configuration of related service strategies.
  • the features associated with the historical behavior of the user to be predicted include at least location information related to the historical behavior to be predicted, and the location information may be a certain city, administrative division, POI, latitude and longitude, etc.
  • the feature associated with the historical behavior of the user to be predicted may include the first feature value under at least one historical behavior feature of the user to be predicted, and the feature of the user to be predicted at the target time within a preset time period in the future.
  • the feature associated with the historical behavior of the user to be predicted may also be one or more historical ride-hailing information of the user to be predicted. The historical ride-hailing information may be obtained based on the user's historical ride-hailing orders.
  • the prediction of the user's location can also be based on the statistics of the user to be predicted, but the prediction accuracy of this method is low, and the location prediction method provided in this application can treat the prediction with higher accuracy.
  • the user’s location is predicted.
  • Fig. 1 is a schematic diagram of an exemplary online platform of an exemplary on-demand service system 100 according to some embodiments of the present application.
  • the on-demand service system 100 may be an online transportation service platform implemented in a network environment, which has a positioning system for providing transportation services.
  • the on-demand service system 100 can provide at least two services. Exemplary on-demand services may include small passenger car sharing services, medium passenger car sharing services, taxi services, driver services, express services, bus services, driver hire services, and shuttle services.
  • the on-demand service system 100 may include a server 110, a network 120, a service requester terminal 130, a service provider terminal 140, a vehicle 150, a storage 160, and a positioning device 170.
  • the server 110 may be a computer server.
  • the server 110 may communicate with the service requester terminal 130 and/or the service provider terminal 140 to provide various functions of online on-demand services.
  • the server 110 may be a single server or a group of servers.
  • the server group may be a centralized server group connected to the network 120 via an access point, or a distributed server group connected to the network 120 via one or more access points.
  • the server 110 may be locally connected to the network 120 or remotely connected to the network 120.
  • the server 110 may access information and/or data stored in the service requester terminal 130, the service provider terminal 140, and/or the storage 160 via the network 120.
  • the storage 160 may be used as a back-end storage of the server 110.
  • the server 110 may be implemented on a cloud platform.
  • the cloud platform may include private clouds, public clouds, hybrid clouds, community clouds, distributed clouds, intermediate clouds, multiple clouds, etc., or any combination thereof.
  • the server 110 may include a processing device 112.
  • the processing device 112 may process information and/or data related to performing one or more functions described in this application.
  • the processing device 112 may obtain historical behavior information of the user.
  • the processing device 112 may determine the first characteristic value, the second characteristic value, and the third characteristic value under the historical behavior characteristic according to the historical behavior information.
  • the processing device 112 may train a position prediction model.
  • the processing device 112 may input the first characteristic value, the second characteristic value, and the third characteristic value under the historical behavior characteristic into the position prediction model, and the prediction model may output the position of the user to be predicted in a preset time period in the future.
  • the processing device 112 may include one or more processing units (for example, a single-core processing engine or a multi-core processing engine).
  • the processing device 112 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), an application specific instruction set processor (ASIP), a graphics processing unit (GPU), a physical processing unit (PPU), and a digital signal processor. (DSP), field programmable gate array (FPGA), programmable logic device (PLD), controller, microcontroller unit, reduced instruction set computer (RISC), microprocessor, etc., or any combination thereof.
  • CPU central processing unit
  • ASIC application specific integrated circuit
  • ASIP application specific instruction set processor
  • GPU graphics processing unit
  • PPU physical processing unit
  • DSP digital signal processor
  • FPGA field programmable gate array
  • PLD programmable logic device
  • controller microcontroller unit
  • RISC reduced instruction set computer
  • the network 120 may facilitate the exchange of information and/or data.
  • one or more components of the on-demand service system 100 can be sent to the on-demand service system 100 via the network 120.
  • Other components in the service system 100 send information and/or data.
  • the server 110 may access and/or obtain at least two pieces of historical behavior information from the storage 160 via the network 120.
  • the network 120 may be any type of wired or wireless network or combination thereof.
  • the network 120 may include a cable network, a wired network, an optical fiber network, a telecommunications network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), and a public telephone. Switched Network (PSTN), Bluetooth network, ZigBee network, Near Field Communication (NFC) network, etc., or any combination thereof.
  • the network 120 may include one or more network access points.
  • the network 120 may include wired or wireless network access points, such as base stations and/or Internet exchange points 120-1, 120-2, and so on.
  • One or more components of the on-demand service system 100 may be connected to the network 120 through a network access point to exchange data and/or information.
  • the passenger or user may be the holder of the service requester terminal 130.
  • the holder of the service requester terminal 130 may be someone other than the passenger.
  • the holder A of the service requester terminal 130 may use the service requester terminal 130 to send a service request for the passenger B, and/or receive service confirmation and/or information or instructions from the server 110.
  • the driver may be a user of the service provider terminal 140.
  • the user of the service provider terminal 140 may be someone other than the driver.
  • the user C of the service provider terminal 140 may use the service provider terminal 140 to receive service requests for the driver D, and/or information or instructions from the server 110.
  • the driver may be designated to use at least one of the service provider terminals 140 and/or 150 types of vehicles for a period of time, for example, one day, one week, one month, or one year. In some other embodiments, the driver may be designated to randomly use one of the service provider terminals 140 and/or one of the vehicles 150. For example, when a driver is available to provide on-demand services, he/she can be assigned to use the driver terminal that receives the earliest request and the vehicle that recommends the on-demand service. In some embodiments, "passenger” and “terminal device” can be used interchangeably, and “driver” and “driver device” can be used interchangeably. In some embodiments, the driver device may be associated with one or more drivers (e.g., night shift drivers, day shift drivers, or randomly moving drivers).
  • drivers e.g., night shift drivers, day shift drivers, or randomly moving drivers.
  • the service requester terminal 130 may include a mobile device 130-1, a tablet computer 130-2, a laptop computer 130-3, a built-in device in a vehicle 130-4, etc., or any combination thereof.
  • the mobile device 130-1 may include a smart home device, a wearable device, a smart mobile device, a virtual reality device, an augmented reality device, etc., or any combination thereof.
  • smart home devices may include smart lighting devices, control devices of smart electrical devices, smart monitoring devices, smart TVs, smart cameras, walkie-talkies, etc., or any combination thereof.
  • the wearable device may include a smart bracelet, smart footwear, smart glasses, smart helmets, smart watches, smart clothing, smart backpacks, smart accessories, etc., or any combination thereof.
  • smart mobile devices may include smart phones, personal digital assistants (PDAs), gaming devices, navigation devices, point of sale (POS) devices, etc., or any combination thereof.
  • the virtual reality device and/or the augmented reality device may include a virtual reality helmet, virtual reality glasses, virtual reality goggles, augmented reality helmets, augmented reality glasses, augmented reality goggles, etc., or any combination thereof.
  • the virtual reality device and/or the augmented reality device may include Google Glass TM , Oculus Rift TM , Hololens TM , Gear VR TM and the like.
  • the built-in devices in the vehicle 130-4 may include a built-in computer, an on-board built-in TV, a built-in tablet computer, and the like.
  • the service requester terminal 130 may include a signal transmitter and a signal receiver, and the signal receiver is configured to communicate with the positioning device 170 to locate the position of the passenger and/or the service requester terminal 130.
  • the service provider terminal 140 may include at least two service provider terminals 140-1, 140-2, ..., 140-n. In some embodiments, the service provider terminal 140 may be similar to or the same as the service requester terminal 130. In some embodiments, the service provider terminal 140 can be customized to implement online transportation services. In some embodiments, the service provider terminal 140 and the service requester terminal 130 may be configured with a signal transmitter and a signal receiver to receive location information of the service provider terminal 140 and the service requester terminal 130 from the positioning device 170. In some embodiments, the service requester terminal 130 and/or the service provider terminal 140 may communicate with other positioning devices to determine the location of the passenger, the service requester terminal 130, the driver, and/or the service provider terminal 140.
  • the service requester terminal 130 and/or the service provider terminal 140 may periodically send the positioning information to the server 110.
  • the service provider terminal 140 may also send the availability status to the server 110 periodically.
  • the availability status may indicate whether the vehicle 150 associated with the service provider terminal 140 is available to transport passengers.
  • the service requester terminal 130 may send positioning information to the server 110 every 30 minutes.
  • the service provider terminal 140 may send the availability status to the server every 30 minutes and/or when the on-demand service is completed.
  • the service requester terminal 130 may send location information to the server 110 every time a user logs in to a mobile application associated with an online on-demand service.
  • the service provider terminal 140 may correspond to one or more vehicles 150.
  • the vehicle 150 can carry passengers and travel to the destination.
  • the vehicle 150 may include at least two vehicles 150-1, 150-2,..., 150-n.
  • One of at least two vehicles may correspond to one order type.
  • Order types can include taxi orders, luxury car orders, express orders, bus orders, shuttle orders, etc.
  • the service may be any online service, such as ordering food, shopping, etc., or a combination thereof.
  • the memory 160 may store data and/or instructions.
  • the memory 160 may store data obtained from the service requester terminal 130 and/or the service provider terminal 140.
  • the memory 160 may store log information associated with the service requester terminal 130.
  • the memory may store bubbling behaviors, billing behaviors, bill-finishing behaviors, etc. that occur through the service requester terminal 130 by the user.
  • the memory 160 may store data and/or instructions that can be executed by the server 110 to provide the on-demand service described in this application.
  • the memory 160 may include mass memory, removable memory, volatile read-write memory, read-only memory (ROM), etc., or any combination thereof. Exemplary mass storage devices may include magnetic disks, optical disks, solid state drives, and the like.
  • Exemplary removable storage may include flash drives, floppy disks, optical disks, memory cards, compact disks, magnetic tapes, and the like.
  • An exemplary volatile read-write memory may include random access memory (RAM).
  • Exemplary RAM may include dynamic RAM (DRAM), double data rate synchronous dynamic RAM (DDR SDRAM), static RAM (SRAM), thyristor RAM (T-RAM), zero capacitor RAM (Z-RAM), and the like.
  • Exemplary ROMs may include mask ROM (MROM), programmable ROM (PROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), compact disk ROM (CD-ROM), and digital Multi-function disk ROM, etc.
  • the storage 160 may be implemented on a cloud platform.
  • the cloud platform may include private clouds, public clouds, hybrid clouds, community clouds, distributed clouds, intermediate clouds, multiple clouds, etc., or any combination thereof.
  • the positioning device 170 may determine information associated with the object, for example, one or more of the service requester terminal 130, the service provider terminal 140, the vehicle 150, and the like. For example, the positioning device 170 may determine the current time and location of the passenger or driver through the service requester terminal 130. In some embodiments, the positioning device 170 may be Global Positioning System (GPS), Global Navigation Satellite System (GLONASS), Compass Navigation System (COMPASS), Beidou Navigation Satellite System, Galileo Positioning System, Quasi-Zenith Satellite System (QZSS) Wait. This information may include the position, height, speed or acceleration of the object, and/or the current time. The position can be in the form of coordinates, such as latitude coordinates and longitude coordinates.
  • the positioning device 170 may include one or more satellites, for example, a satellite 170-1, a satellite 170-2, and a satellite 170-3.
  • the satellites 170-1 to 170-3 can determine the above-mentioned information independently or collectively.
  • the positioning system 170 may transmit the above-mentioned information to the service requester terminal 130, the service provider terminal 140, or the vehicle 150 via the network 120.
  • one or more components in the on-demand service system 100 can access data or instructions stored in the memory 160 via the network 120.
  • the storage 160 may be directly connected to the server 110 as a back-end storage.
  • one or more components in the on-demand service system 100 may have permission to access the memory 160.
  • one or more components in the on-demand service system 100 can read and/or modify information related to passengers, drivers, and/or vehicles.
  • the server 110 may read and/or modify the user characteristics of one or more passengers after completing the on-demand service order.
  • the information exchange between one or more components of the on-demand service system 100 can be initiated by launching a mobile application of the on-demand service on the terminal device to request the service.
  • the object of the service request can be any product.
  • the product may include food, medicine, commodities, chemical products, electrical appliances, clothes, automobiles, household goods, luxury goods, etc., or any combination thereof.
  • products may include service products, financial products, knowledge products, Internet products, etc., or any combination thereof.
  • Internet products may include personal host products, network products, mobile Internet products, commercial host products, embedded products, etc., or any combination thereof.
  • Mobile Internet products can be used in mobile terminals, programs, systems, etc. or any combination of software.
  • the mobile terminal may include a tablet computer, a laptop computer, a mobile phone, a personal digital assistant (PDA), a smart watch, a point of sale (POS) device, a vehicle-mounted computer, a vehicle-mounted TV, a wearable device, etc., or any combination thereof.
  • the product can be any software and/or application used in a computer or mobile phone.
  • the software and/or application program may involve social networking, shopping, transportation, entertainment, learning, investment, etc., or any combination thereof.
  • transportation-related software and/or applications may include travel software and/or applications, vehicle scheduling software and/or applications, mapping software and/or applications, and the like.
  • vehicles can include horses, carriages, rickshaws (for example, unicycles, bicycles, tricycles, etc.), automobiles (for example, taxis, buses, private cars, etc.), trains, subways, boats, and aircraft (For example, airplane, helicopter, space shuttle, rocket, hot air balloon, etc.), etc., or any combination thereof.
  • rickshaws for example, unicycles, bicycles, tricycles, etc.
  • automobiles for example, taxis, buses, private cars, etc.
  • trains subways, boats, and aircraft (For example, airplane, helicopter, space shuttle, rocket, hot air balloon, etc.), etc., or any combination thereof.
  • the element when an element of the on-demand service system 100 operates, the element may operate through an electrical signal and/or an electromagnetic signal.
  • the service requester terminal 130 may operate a logic circuit in its processor to process such a task.
  • the processor of the service requester terminal 130 may generate an electrical signal encoding the order.
  • the processor of the service requester terminal 130 may transmit the electric signal to the output port. If the service requester terminal 130 communicates with the server 110 via a wired network, the output port may be physically connected to a cable, and the cable also transmits an electric signal to the input port of the server 110.
  • the output port of the terminal 130 may be one or more antennas, which convert electrical signals into electromagnetic signals.
  • the service provider terminal 140 may process tasks through the operation of logic circuits in its processor, and receive instructions and/or service commands from the server 110 via electrical or electromagnetic signals.
  • the processor processes instructions, issues instructions, and/or executes actions, the instructions and/or actions are performed through electrical signals.
  • the processor when the processor retrieves data (for example, the user's historical order information, the user's bubbling behavior information, etc.) from the storage medium (for example, the memory 160), it may send an electrical signal to the reading device of the storage medium, which Can read structured data in storage media.
  • the structured data can be sent to the processor in the form of electrical signals via the bus of the electronic device.
  • the electrical signal may refer to one electrical signal, a series of electrical signals, and/or at least two discrete electrical signals.
  • FIG. 2A is a schematic diagram of exemplary hardware and/or software components of a computing device 200A according to some embodiments of the present application.
  • the computing device 200A may be a general-purpose computer or a special-purpose computer. Both can be used to realize the functions of this application.
  • the computing device 200A can be used to implement any component of the service as described herein.
  • the processing device 112 of the server may be implemented on the computing device 200A through its hardware, software program, firmware, or a combination thereof.
  • the service-related computer functions described herein can be implemented in a distributed manner on at least two similar platforms to distribute the processing load.
  • the computing device 200A may include a communication port 250a connected to a network (e.g., network 120) connected to it to facilitate data communication.
  • the computing device 200A may also include one or more processors 220a for executing program instructions.
  • An exemplary computer platform may include an internal communication bus 210a, various forms of program memory and data storage for various data files processed and/or transmitted by the computer, such as a disk 270a, a read only memory (ROM) 230a, or a random access memory ( RAM) 240a.
  • the exemplary computer platform may also include program instructions stored in ROM 230a, RAM 240a, and/or other types of non-transitory storage media to be executed by CPU 220a.
  • the methods and/or processes of the present application can be implemented as program instructions.
  • the computing device 200A also includes an input/output (I/O) component 260a, which supports input/output between the computer, the user, and other components herein.
  • the computing device 200A can also receive programs and data through
  • the computing device 200A in this application may also include multiple CPUs and/or processors, so the operations and/or method steps performed by one CPU and/or processor described in this application may also be Multiple CPUs and/or processors execute jointly or individually.
  • the CPU and/or the processor of the computing device 200A may perform step A and step B.
  • step A and step B can also be performed jointly or separately by two different CPUs and/or processors in the computing device 200A (for example, the first processor performs step A and the second processor performs Step B, or the first and second processors jointly perform steps A and B).
  • FIG. 2B is a schematic diagram of exemplary hardware and/or software components of a mobile device 200B according to some embodiments of the present application.
  • the mobile device 200B may include a communication module 210b, a display 220b, a graphics processing unit (GPU) 230b, a central processing unit (CPU) 240b, an input/output (I/O) 250b, a memory 260b, and a memory 290b.
  • any other suitable components including but not limited to a system bus or a controller (not shown), may also be included in the mobile device 200B.
  • the mobile operating system 270b eg, iOS TM , Android TM , Windows Phone TM, etc.
  • one or more application programs 280b may be loaded from the memory 290b into the memory 260b so as to be executed by the CPU 240b.
  • the application program 280b may include a browser or any other suitable mobile application program for sending, receiving, and presenting information related to on-demand services (for example, sending car-hailing orders) from the processing device 112 and/or the memory 160.
  • the interaction between the user and the information flow can be implemented through the I/O 250b, and provided to the processing device 112 and/or other components of the on-demand service system 100 through the network 120.
  • Fig. 3 shows a flowchart of an exemplary process for predicting a position according to some embodiments of the present application.
  • the process 300 may include steps S301 to S302.
  • the process 300 may be performed by the processing device 112.
  • the process 300 may be implemented with a set of instructions (for example, an application program) stored in the memory 160.
  • the server 110, the CPU 220a, and/or the CPU 240b can execute the set of instructions, and thus can instruct to execute the process 300.
  • S301 Obtain the first feature value of the user to be predicted under at least one historical behavior feature, and the second feature value of the user to be predicted under the target time feature and the second feature value under the target location feature within a preset time period in the future.
  • the third characteristic value Obtain the first feature value of the user to be predicted under at least one historical behavior feature, and the second feature value of the user to be predicted under the target time feature and the second feature value under the target location feature within a preset time period in the future.
  • the users to be predicted may be all users of an online transportation service platform (for example, a car-hailing platform).
  • an online transportation service platform for example, a car-hailing platform.
  • the location of each user is determined, and the amount of calculation is very large. Therefore, in order to be more targeted, It can also be a user selected from all users of the online transportation service platform based on certain conditions. For example, the location of each user during bubbling can be calculated periodically, and if the bubbling location of any user during the period belongs to at least two cities or a preset area, the user is determined as the user to be predicted. For another example, it is possible to periodically calculate the place of departure from which each user places an order.
  • the user is determined to be a waiting area.
  • Forecast users For another example, it may be determined that a user whose historical order quantity is greater than the threshold order quantity in a threshold time period is a user to be predicted.
  • the threshold time period can be the past day, week, month, year, etc.
  • the methods of obtaining the third characteristic value are described separately:
  • FIG. 4 shows a flowchart of a process of obtaining a first characteristic value under a first historical behavior characteristic and a first characteristic value under a second historical behavior characteristic according to some embodiments of the present application.
  • the process 400 may include steps S401 to S403.
  • the process 400 may be performed by the processing device 112.
  • the process 400 may be implemented with a set of instructions (for example, an application program) stored in the memory 160.
  • the server 110, the CPU 220a, and/or the CPU 240b can execute the set of instructions, and therefore can instruct to execute the process 400.
  • S401 Extracting regional historical behavior information of the user to be predicted in different regions (including but not limited to the region to be predicted) from at least one type of historical behavior information of the user to be predicted, and extracting multiple historical behaviors of the user to be predicted Regional time historical behavior information in different regions within each historical time period in the time period.
  • the area may be a province, city, county, town, district, etc., in a municipal plan.
  • the area may be a grid divided according to latitude and longitude.
  • the area may be an area in which longitude and latitude are between 90 degrees to 91 degrees east longitude and 30 degrees to 31 degrees north latitude, respectively.
  • the area may also be an arbitrarily divided area of interest (AOI).
  • AOI area of interest
  • the area may be a school, a shopping mall, a scenic spot, a hospital, etc.
  • the area may be a user-defined area. For example, an area with a certain point as the center and a certain distance as a radius.
  • the historical behavior of the user to be predicted may include any operation behavior of the user in the online transportation service platform.
  • the historical behavior may include login behavior, bubbling behavior, billing behavior, order completion behavior, evaluation behavior, consulting behavior, subscription behavior, etc., or any combination thereof.
  • the historical behavior may be a historical behavior in a certain period of time in the past (for example, the previous day, the previous week, the previous month, the last year, etc.). Each type of historical behavior can correspond to a type of historical behavior information.
  • bubbling means that the service requester enters the home page of the service software of the online transportation service platform, and sends the travel starting point and travel destination of the service requesting end to the online transportation service platform.
  • the online transportation service platform receives the travel starting point and travel end point sent by the service requester, and determines to monitor bubbling behavior on the service requesting end.
  • the user can log in to the service software through the user terminal (for example, the service requester terminal 130), the user terminal can upload the location information of the user terminal through the positioning system (for example, the positioning device 170) as the starting point of travel, and the user can manually input or drag
  • the travel destination is input by dragging the map.
  • the on-demand service system 100 may store the user's behavior as a bubbling behavior and store it in the memory 160.
  • the online transportation service platform After the online transportation service platform obtains the travel starting point and travel end point sent by the service requesting terminal, it determines the estimated order price for the service requesting terminal for this trip, and sends the estimated order price to the service requesting terminal. After receiving the estimated order price, the service requester generates order information based on the passenger's trigger, and sends the order information to the online transportation service platform. This behavior is the billing behavior of the service requester.
  • the online transportation service platform After the online transportation service platform receives the order information sent by the service requester, it generates an order and pushes the order to multiple qualified service providers, and matches the service provider for the service requester based on the feedback of the service provider.
  • the service provider matched with the service requester After the service provider matched with the service requester completes the service, it will feedback the service completion information to the online transportation service platform.
  • the online transportation service platform initiates fee payment to the service requester. After the service requester completes the payment, it is deemed that the service requester has completed the order.
  • the billing behavior is not the same as the billing behavior. Only when the order is completed and the payment is completed, the user behavior is considered to be the order completion behavior. If in the middle of the service process, the service provider, service requester, or online ride-hailing party cancels the service, the order completion behavior corresponding to the order will no longer occur.
  • Each historical behavior can include corresponding historical behavior information.
  • the historical behavior information corresponding to the bubbling behavior includes historical bubbling behavior information;
  • the historical behavior information corresponding to the billing behavior includes historical billing behavior information;
  • the historical behavior information corresponding to the billing behavior includes historical billing behavior information.
  • Each type of historical behavior information may include the time when the historical behavior occurred and the place where the historical behavior occurred. For example, when a user has a historical behavior through the user terminal, the user terminal may send the historical behavior occurrence time and the historical behavior occurrence location when the historical behavior occurred to the memory 160 for storage. As described here, the time when the historical behavior occurred and the place where the historical behavior occurred respectively refer to the time and place when the historical behavior was completed. For example, a user can log in to the service software.
  • the online transportation service platform can record the user's behavior as a bubbling behavior, and input the time and place of the trip start and/or end of the trip as a historical behavior Occurrence time and location of the historical behavior; in response to the user triggering the generation of order information, the online transportation service platform can update the user’s behavior as an ordering behavior, and determine the time and place that triggered the generation of the order information as the historical behavior time and history Where the act occurred.
  • the at least one type of historical behavior information of the user to be predicted includes historical behavior information of at least one of the foregoing historical behaviors.
  • the following methods may be used to obtain at least one type of historical behavior information, regional historical behavior information, and regional time historical behavior information of the user to be predicted.
  • the i-th historical behavior information is recorded as x i .
  • the historical behavior information X of the user to be predicted may be obtained from the memory 160 or other storage devices (not shown).
  • each historical behavior information X determines the regional historical behavior information X city of the user to be predicted in different regions.
  • the j-th regional historical behavior information is recorded as .
  • some or all of the different regions may be regions to be predicted.
  • the regional historical behavior information of the user to be predicted in different regions the historical behavior occurrence time of X city and the preset multiple historical time periods, it is determined that the user to be predicted is in a different historical time period in each of the multiple historical time periods.
  • the regional time history behavior information of the region X city,t The number of groups of the regional time historical behavior information X city, t may be equal to the number of segments N of the multiple historical time periods ⁇ the number of regions M.
  • the n-th regional historical time behavior information of the m-th group is recorded as
  • multiple historical time periods can be specifically set according to actual needs.
  • multiple historical time periods can be divided according to the distance from the current time, for example, 0-2 days from the current, 2-5 days from the current, and The current 5-10 days, the current 10-20 days, the current 20-50 days, the current 50-100 days, and the current 100 days or more.
  • the corresponding regional historical time behavior information can be 7 groups.
  • S402 Determine the first feature under the first historical behavior feature of the user to be predicted according to the at least one type of historical behavior information of the user to be predicted, the regional historical behavior information, and the regional time historical behavior information value.
  • S403 Determine a first characteristic value under a second historical behavior characteristic of the user to be predicted according to the at least one historical behavior information of the user to be predicted.
  • the first historical behavior feature includes but is not limited to one or more of the following a1 to a5:
  • a1 The number of occurrences of historical behaviors corresponding to multiple points of interest (POI) categories.
  • POI classification is a classification determined in advance for different POIs. For example: airports, entertainment venues, shopping malls, schools, etc.
  • the number of occurrences of historical behaviors per day in a preset time period For example, the number of occurrences of historical behavior from Monday to Sunday.
  • the number of occurrences of historical behavior on working days refers to the number of bubbling behaviors and billing behaviors on working days.
  • the corresponding occurrence times of different historical behaviors include the corresponding occurrence times of bubbling behavior and the occurrence times corresponding to billing behavior.
  • the second historical behavior feature includes but is not limited to one or more of the following b1 to b5:
  • b1 Whether the area where the last historical behavior occurred is the area; where the last historical behavior refers to the current time (the current time can refer to any preset time, such as model training time, model prediction time or other specified Time) the most recent historical act.
  • the time interval between the time when the last historical behavior occurred and the time to be predicted for example, if the time when the last historical behavior occurred is 2 days from the current time, and the time to be predicted is 1 day in the future, then the time interval For 3 days. For another example, if the occurrence time of the last historical behavior is 5 hours from the current time, and the time to be predicted is 5 to 8 hours in the future, the time interval is 10 hours.
  • the number of areas reached by the users to be predicted refers to the area determined by the on-demand service system 100 to locate the user terminal through the positioning device 170 after the user logs in to the service software through the user terminal (for example, the service requester terminal 130). For example, if the on-demand service system 100 uses the positioning device 170 to locate the user who has logged in to the service software in Chengdu, it can determine that the user has arrived in Chengdu.
  • the area and the time to be predicted may be the default settings of the on-demand service system 100, or may be specifically set according to actual application conditions.
  • the area may be an area where the user to be predicted has reached. In some embodiments, the area may also be an area that the user to be predicted has not reached. For example, the area may be a destination to be predicted that the user has searched for.
  • the time to be predicted is a preset time period in the future.
  • the location prediction method provided by the embodiment of the present application can predict the probability that the user to be predicted will appear in the area in a preset time period in the future.
  • the first feature value of the user to be predicted under at least one historical behavior feature may include three parts:
  • the first part According to at least one type of historical behavior information X of the user to be predicted, a feature value under the first historical behavior feature corresponding to the at least one type of historical behavior information X of the user to be predicted is determined.
  • the predetermined time period in the future is the time to be predicted.
  • the second feature value under the target time feature may reflect information related to the time to be predicted.
  • the target time feature may include whether the future preset time period is the day of the week, and whether the future preset time period is a holiday or a working day. In some embodiments, the target time feature may also include the attribute of the future preset time period. For example, whether it is in peak hours or peak off hours.
  • the target time feature can be determined, so that the second feature value of the user to be predicted under the target time feature in a preset time period in the future can be determined.
  • the second feature value may be a digitization of the target time feature.
  • the second feature value may be a vector obtained by encoding the target time feature.
  • the target location feature may include the attributes of the area. For example, whether the area is a tourist city, whether it is a national capital city, the POI classification type near the area, and the distance between the area and the user's home and/or company to be predicted.
  • the corresponding third characteristic value can be determined according to the attributes of the area.
  • the third feature value may be the digitization of the target location feature.
  • the third feature value may be a vector obtained by encoding the target location feature.
  • the process 300 further includes:
  • S302 Input the first feature value, the second feature value, and the third feature value into a pre-trained location prediction model, and obtain the location prediction of the user to be predicted in the future preset time period result.
  • the location prediction result of the user to be predicted in the predetermined time period in the future may be the probability that the user to be predicted will appear at the location to be predicted in the predetermined time period in the future.
  • the probability that the user to be predicted will appear at multiple locations to be predicted within a preset time period in the future can also be predicted.
  • the third feature value corresponding to each area of the user to be predicted is also obtained, and the input of multiple groups of location prediction models is formed , Respectively input into the location prediction model to obtain the probability that the user to be predicted will appear in each area at a preset time period in the future.
  • the on-demand service system 100 may determine the area with the greatest probability as the location where the user to be predicted will appear in the future preset time period.
  • the first feature value a of the user to be predicted the second feature value b under the target time feature in the future preset time period, and obtain the user to be predicted in the future preset time period, and the area A, B and C correspond to the third characteristic values c1, c2, and c3, respectively.
  • three sets of input data are formed: a, b, and c1; a, b, and c2; a, b, and c3.
  • the three sets of input data are sequentially input into the pre-trained location prediction model to obtain the probabilities of the users to be predicted in places A, B, and C in the next twenty-four hours.
  • the user to be predicted after determining the probability that the user to be predicted appears in multiple regions, it can be sorted according to the probability of the user to be predicted appearing in the multiple regions, and the region with the highest probability can be determined as the user to be predicted in the future. Set the most likely location within the time period, so that resource scheduling and related service strategy configuration can be performed in advance based on the prediction result.
  • Fig. 5 shows a flowchart of a process of training a position prediction model according to some embodiments of the present application.
  • the process 500 may include steps S501 to S502.
  • the process 500 may be performed by the processing device 112.
  • the process 500 may be implemented with a set of instructions (for example, an application program) stored in the memory 160.
  • the server 110, the CPU 220a, and/or the CPU 240b can execute the set of instructions, and therefore can instruct to execute the process 500.
  • the process 500 may be performed by other systems or devices other than the on-demand service system 100. For example, implemented by equipment or systems provided by a manufacturing supplier.
  • S501 Acquire a first sample feature value of a plurality of sample users under at least one historical behavior characteristic, and a second sample feature value of each of the sample users under a target time characteristic within a historical preset time period and The third sample feature value under the target location feature; the sample users include positive sample users and negative sample users.
  • S502 Train the position prediction model based on the first sample feature value, the second sample feature value, and the third sample feature value of each sample user.
  • the position prediction model may include xgboost (eXtreme Gradient Boosting), Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), and Decision Tree (DT). ) Model, gradient boosting decision tree (Gradient Boosting Decision Tree, GBDT) model, k-nearest neighbor algorithm (K-Nearest Neighbor, kNN) model, convolutional neural network (Convolutional Neural Networks, CNN), artificial neural network (Artificial Neural) Networks, ANN) models, etc., or any combination thereof.
  • xgboost eXtreme Gradient Boosting
  • RNN Recurrent Neural Network
  • LSTM Long Short-Term Memory
  • DT Decision Tree
  • the historical preset time periods of different sample users may be the same or different. For example, sample user A determines "January 15, 2018” as the historical preset time period, and sample user B determines "October 10, 2018” as the historical preset time period. However, the duration of the historical preset time period of different sample users is the same, and the duration of the aforementioned future preset time period is the same.
  • the sample user appears in the area during the historical preset time period, the sample user is a positive sample user. If the sample user does not appear in the area in the historical preset time period, the sample user is a negative sample user.
  • the probability of their appearance in the area during the historical preset time period can be set to 1; the negative sample user does not appear in the historical preset time period In the predicted area, the probability of appearing in the area in the historical preset time period can be set to zero.
  • the process of training the model After the first sample feature value, second sample feature value, and third sample feature value of each sample user are input into the location prediction model, the output result of the location prediction model should be as close as possible to the The probability of the area.
  • the location prediction model can only predict the probability that the user to be predicted will appear in a certain area within a preset time period in the future, or it may be for the user to be predicted to appear in multiple areas to be preset within a preset time period in the future. Therefore, for these two different situations, the situation of the determined sample users is also different.
  • the location prediction model can only predict the probability of the user to be predicted in a certain area within a preset time period in the future, the following methods can be used to determine the sample users:
  • Filters users related to the area may include users who have visited the area and/or users who have searched the area.
  • the following method can be used to determine the first sample feature value of the sample user:
  • For each sample user extract regional historical behavior information of the sample user in the region from at least one type of historical behavior information of the sample user, and extract each historical time period of the sample user in a plurality of historical time periods Inside, the regional time history behavior information of the area to be tested;
  • a first sample characteristic value under the second historical behavior characteristic of the sample user is determined.
  • the first sample feature value under the first historical behavior feature of the sample user and the first sample feature value under the second historical behavior feature of the sample user are the same as the method for determining the first feature value of the user to be predicted Similar, I won't repeat it here.
  • the first sample feature value of the sample user is constructed based on historical behavior information before the historical preset time period.
  • the location prediction model can predict the probability of the user to be predicted in multiple areas within a preset time period in the future, the following methods can be used to determine the sample users:
  • the following methods can be used to obtain the first sample feature value, the second sample feature value, and the third sample feature value of the sample user:
  • any one of the following methods (1) and (2) may be used to obtain the first sample feature value of each sample user:
  • a first sample characteristic value under the second historical behavior characteristic of the sample user is determined.
  • the regional historical behavior information of the user in at least one region and extracting the regional time historical behavior information corresponding to each region in at least one region in each historical time period of the multiple historical time periods, it is determined that the user is in At least one area corresponds to the first sample feature value of each area under the first historical behavior feature;
  • a first sample feature value under the second historical behavior feature corresponding to each area of the user in at least one area is determined.
  • the regional historical behavior information corresponding to the region M1 can be extracted, and the regional time historical behavior information corresponding to the region M1 can be extracted.
  • a first sample feature value corresponding to the user C and the area M1 is constructed.
  • a first sample feature value corresponding to the user C and the area M2 is constructed.
  • the first sample feature value under the first historical behavior feature of the sample user and the first sample feature value under the second historical behavior feature of the sample user are the same as the method for determining the first feature value of the user to be predicted Similar, I won't repeat it here.
  • sample users in different regions can be the same or different. For example, if user A has appeared in area A and area B, then the user A can be a sample user corresponding to area A or a sample user corresponding to area B.
  • the historical behavior information of A in area A and B is different, when A is a sample user in area A and B, the first sample feature value corresponding to area A is generated, and The feature value of the first sample corresponding to the B area is different. The corresponding characteristic value of the third sample is also different.
  • the position prediction model can be trained in the following way:
  • the user for example only, the user’s historical behavior related data (for example, the first feature value, the second feature value, and the third feature value) can be used as input, and the corresponding analysis result can be used as the output and the user appears in the historical preset time period.
  • the correct standard Ground Truth
  • the model parameters can be adjusted inversely according to the direct difference between the predicted output of the model (for example, the predicted position) and the correct standard.
  • the training process will stop .
  • a certain preset condition for example, the training sample reaches a predetermined number, the prediction accuracy of the model is greater than a predetermined accuracy threshold, or the value of the loss function (Loss Function) is less than a preset value, the training process will stop .
  • the loss function Loss Function
  • the embodiment of the present application also provides a specific method for training the position prediction model based on the first sample feature value and the second sample feature value of each sample user, including: The first sample feature value, the second sample feature value, and the third sample feature value are constructed to construct multiple sub-decision trees; based on the multiple sub-decision trees, it is determined as the position prediction model.
  • Fig. 6 shows a flowchart of a process of training a position prediction model according to some embodiments of the present application.
  • the process 600 may include steps S601 to S605.
  • the process 600 may be performed by the processing device 112.
  • the process 600 may be implemented with a set of instructions (for example, an application program) stored in the memory 160.
  • the server 110, the CPU 220a, and/or the CPU 240b can execute the set of instructions, and therefore can instruct to execute the process 600.
  • the process 600 may be performed by other systems or devices other than the on-demand service system 100. For example, implemented by equipment or systems provided by a manufacturing supplier.
  • S601 randomly determine multiple target characteristics from the historical behavior characteristics, the target time characteristics, and the target area characteristics;
  • S602 Construct a sub-decision tree of the current iteration cycle based on the feature value of the sample user under each of the target features;
  • S603 Based on the sub-decision tree of the current iteration period and the sub-decision tree of the historical iteration period, constitute the current decision tree set, and determine the loss of the current decision tree set;
  • S604 Detect whether the loss of the current decision tree set is greater than the preset loss threshold; if so, jump to S601; the current iteration cycle is completed. If not, skip to S605.
  • S605 Determine the current decision tree set as the position prediction model.
  • multiple target characteristics are randomly determined, based on the multiple randomly determined target characteristics and the characteristic value of each sample user under the corresponding target characteristic, and the relationship with the sample user
  • the corresponding probability of appearing at the position to be predicted is trained on the position prediction model to obtain the decision tree set.
  • the target feature is adjusted according to the loss of the aforementioned decision tree set, and the selection probability prediction model is trained according to the newly re-determined target feature to obtain the current decision tree set.
  • the user verifies the current decision tree set based on the test sample, and determines the loss of the current decision tree set.
  • the location prediction model predicts the location of the user to be predicted, it inputs the first feature value, the second feature value, and the third feature value into a pre-trained location prediction model to obtain each Sub-prediction results of the decision tree;
  • the embodiment of the application obtains the first characteristic value of the user to be predicted under at least one historical behavior characteristic, and the second characteristic value of the user to be predicted under the target time characteristic and the target location characteristic within a preset time period in the future. And input the first feature value, the second feature value, and the third feature value into the pre-trained location prediction model to obtain the location prediction of the user to be predicted in the future preset time period.
  • the pre-trained location prediction model to obtain the location prediction of the user to be predicted in the future preset time period.
  • FIG. 7A shows a flowchart of an exemplary process for predicting a position according to some embodiments of the present application.
  • the process 700A may include steps S701 to S703.
  • the process 700A may be performed by the processing device 112.
  • the process 700A may be implemented with a set of instructions (eg, application programs) stored in the memory 160.
  • the server 110, the CPU 220a, and/or the CPU 240b can execute the set of instructions, and therefore can instruct to execute the process 700A.
  • S701 Obtain the first feature value of the user to be predicted under at least one historical behavior feature, and the second feature value of the user to be predicted under the target time feature and the second feature value under the target location feature within a preset time period in the future The third characteristic value.
  • S701 is similar to the implementation of S301 described above, and will not be repeated here.
  • S702 Obtain a fourth characteristic value of the user to be predicted under multiple user attribute characteristics.
  • the user attribute characteristics include one or more of the following:
  • the number of historical orders of the user whether it is a business person, whether it is a tourist, the area where the user’s home is located, and the area where the user’s company is located.
  • S703 Input the first feature value, the second feature value, the third feature value, and the fourth feature value into a pre-trained location prediction model, and obtain the user to be predicted in the future The location prediction result within a preset time period.
  • S703 is similar to the implementation of S302 described above. I won't repeat them here.
  • the fourth sample feature value of each of the sample users under multiple user attribute features can also be obtained; when the location prediction model is trained, Training the position prediction model based on the first sample feature value, the second sample feature value, the third sample feature value, and the fourth sample feature value of each sample user.
  • the specific training method is similar to the model training method in S302, and will not be repeated here.
  • the historical ride-hailing information of the user to be predicted, the preset time period in the future, and the area to be predicted may be processed separately with user attribute characteristics and then input into the trained location prediction model to determine the Predict the probability that the user will be in the area within a preset time period in the future.
  • the historical taxi information may include the origin, destination, attributes of both, and time information.
  • the attributes can include the city, area, or POI information of the origin or destination.
  • the time information may include the time of issuing the order, the time of departure or the time of arrival, etc.
  • several historical ride-hailing information before the time to be predicted may be input into the position prediction model, such as historical ride-hailing information 100 times, 200 times, 500 times before the time to be predicted, and so on.
  • the position prediction model can be obtained by training the initial neural network model.
  • the initial neural network model may include xgboost model, recurrent neural network, long short-term memory network, k-nearest neighbor (kNN) model, convolutional neural network (Convolutional Neural Networks, CNN) , Artificial Neural Networks (ANN) models, etc., or any combination thereof.
  • Fig. 7B shows a flowchart of a prediction process of a position prediction model according to some embodiments of the present application. As shown in FIG. 7B, the historical ride-hailing information of the user to be predicted, user attribute characteristics, future preset time period, and the area to be predicted can be obtained, and then input into the location prediction model.
  • the vector xt and vector q are processed based on the attention mechanism to obtain the vector e.
  • the position prediction model may forward the vector e, and obtain y after two-layer neural network and activation function processing. The y may reflect the arrival of the user to be predicted at the location to be measured within a preset time period in the future. Probability.
  • training data may be obtained to train the initial neural network model to obtain a position prediction model.
  • the training data may include historical ride-hailing information, target time period, target area, user attribute characteristics, and label data of multiple sample users.
  • the label data reflects whether the sample user has reached the target area during the corresponding target time period. For example, if the sample user has reached the target area during the corresponding target time period, the label can be 1, otherwise, it is 0. For different sample users, the corresponding target time and target area may be different.
  • the prediction result of the model may be a probability value, which reflects the predicted probability of the sample user going to the target area within the target time period.
  • the loss function can be constructed based on the difference between the label data of the training data and the model prediction result, and the parameters of the model can be adjusted based on the loss function. When the value of the loss function is less than a certain preset value, the training process will stop. Otherwise, update the parameters of the initial neural network model and recalculate until the value of the loss function is less than the preset value.
  • the position prediction end-to-end model can be updated periodically or in real time according to the update of the user's behavior on the online transportation service platform. For example, when some users in the platform complete a new order, the on-demand service system 100 can update the user's historical taxi information based on the new order completion behavior, and then generate new training samples based on the updated historical taxi information, and then update the location Forecast end-to-end models.
  • Fig. 8 shows an exemplary block diagram of a position prediction apparatus according to some embodiments of the present application.
  • the position prediction device may include an acquisition module 81, a prediction module 82 and a training module 83.
  • the acquiring module 81 is configured to acquire the first characteristic value of the user to be predicted under at least one historical behavior characteristic, and the second characteristic value of the user to be predicted under the target time characteristic and the second characteristic value of the user to be predicted in the future preset time period.
  • the prediction module 82 is configured to input the first feature value, the second feature value, and the third feature value into a pre-trained position prediction model to obtain the preset time of the user to be predicted in the future The position prediction result of the segment.
  • the acquiring module 81 is configured to acquire the first characteristic value of the user to be predicted under at least one historical behavior characteristic in the following manner:
  • the first feature value under the second historical behavior feature of the user to be predicted is determined.
  • the first historical behavior feature includes one or more of the following:
  • the number of occurrences of historical behaviors corresponding to multiple POI classifications of points of interest The number of occurrences of historical behaviors per day within a preset time period, the number of occurrences of historical behaviors on working days, the number of occurrences of historical behaviors on non-working days, and different histories The number of occurrences corresponding to each behavior.
  • the second historical behavior feature includes one or more of the following:
  • the region where the last historical behavior occurred is the region, the time interval between the time when the last historical behavior occurred and the time to be predicted, the POI classification of the destination of the last historical behavior, and the departure place of the last historical behavior
  • the historical behavior information includes one or more of historical bubbling behavior information, historical billing behavior information, and historical billing behavior information.
  • the acquiring module 81 is further configured to acquire the fourth characteristic value of the user to be predicted under multiple user attribute characteristics
  • the prediction module 82 is configured to obtain the position prediction result of the user to be predicted in the future preset time period in the following manner:
  • the inputting the first feature value, the second feature value, the third feature value, and the fourth feature value into a pre-trained position prediction model, and obtaining the user to be predicted in the future The location prediction result within a preset time period.
  • the user attribute characteristics include one or more of the following: the number of orders, whether it is a business person, whether it is a tourist, the area where the user's home is located, and the area where the user's company is located.
  • the training module 83 is used to train the position prediction model using the following device:
  • the sample users include positive sample users and negative sample users;
  • the training module 83 is configured to obtain the first sample feature value of a plurality of sample users under at least one historical behavior characteristic in the following manner, and the historical preset time period of each sample user , The second sample feature value under the target time feature and the third sample feature value under the target location feature:
  • the training module 83 is configured to train the position prediction model based on the first sample feature value, the second sample feature value, and the third sample feature value of each sample user in the following manner:
  • the training module 83 is configured to train the position prediction model based on the first sample feature value and the second sample feature value of each sample user in the following manner:
  • a plurality of the sub-decision trees are determined as the position prediction model.
  • the training module 83 is configured to construct the first sample feature value, the second sample feature value, and the third sample feature value of each sample user in the following manner Multiple sub-decision trees:
  • the loss is greater than the preset loss threshold, complete the current iteration cycle and return to the historical behavior characteristics, the target time characteristics, the target area characteristics, and the user attribute characteristics. Steps of target characteristics;
  • the current decision tree set is determined as the position prediction model.
  • the training module 83 is used to determine the loss of the current decision tree set in the following manner:
  • the first test sample feature value of multiple test sample users under the historical behavior characteristics, and the second test sample feature value of each test sample user under the to-be-predicted time feature and the regional feature are input to all In the current decision tree set, obtain a location prediction result corresponding to each of the test sample users;
  • the loss of the current decision tree set is determined.
  • the training module 83 is further configured to obtain the fourth sample feature value of each sample user under multiple user attribute features
  • the training module 83 is configured to train the position prediction model based on the first sample feature value, the second sample feature value, and the third sample feature value of each sample user in the following manner:
  • the prediction module 82 is configured to input the first feature value, the second feature value, and the third feature value into a pre-trained position prediction model in the following manner to obtain The location prediction result of the user to be predicted in the future preset time period:
  • the aforementioned modules may be connected or communicate with each other via a wired connection or a wireless connection.
  • Wired connections may include metal cables, optical cables, hybrid cables, etc., or any combination thereof.
  • the wireless connection may include a connection in the form of LAN, WAN, Bluetooth, ZigBee, or NFC, or any combination thereof. Two or more modules can be combined into a single module, and any one module can be divided into two or more units.
  • an embodiment of the present application also provides an electronic device, including: a bus 210a, a processor 220a, a communication port 250a, an input/output 260a, and a storage medium (for example, a magnetic disk 270a, a read-only memory (ROM) 230a , Or random access memory (RAM) 240a), the storage medium stores machine-readable instructions executable by the processor 220a, and when the electronic device is running, the processor 220a communicates with the storage medium through a bus 210a.
  • the processor 220a executes the machine-readable instructions to execute the steps of the position prediction method provided in the embodiments of the present application when executed.
  • the embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored, and the computer program executes the steps of the position prediction method provided in the embodiment of the present application when the computer program is run by a processor.
  • the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • the functional units in the various embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a nonvolatile computer readable storage medium executable by a processor.
  • the technical solution of this application essentially or the part that contributes to the existing technology or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in each embodiment of the present application.
  • the aforementioned storage media include: U disk, mobile hard disk, ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

Abstract

La présente invention concerne un procédé et un dispositif de prédiction d'emplacement. Le procédé consiste : à acquérir une caractéristique associée à un comportement historique d'un utilisateur dont l'emplacement doit être prédit, la caractéristique comprenant au moins des informations d'emplacement concernant le comportement historique de l'utilisateur ; et à utiliser un modèle de prédiction d'emplacement pré-formé afin d'acquérir, au moins sur la base de la caractéristique associée au comportement historique de l'utilisateur, un résultat d'un emplacement prédit de l'utilisateur dans une période de temps future prédéfinie. Le procédé permet la détermination d'un emplacement d'un utilisateur dans une période de temps future prédéfinie, de telle sorte que des informations d'emplacement de l'utilisateur peuvent être prédites à l'avance, ce qui facilite la planification de ressources à l'avance et la configuration de schémas de service associés.
PCT/CN2020/073652 2019-01-21 2020-01-21 Procédé et dispositif de prédiction d'emplacement WO2020151725A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910055395.8A CN111461380A (zh) 2019-01-21 2019-01-21 一种位置预测方法以及装置
CN201910055395.8 2019-01-21

Publications (1)

Publication Number Publication Date
WO2020151725A1 true WO2020151725A1 (fr) 2020-07-30

Family

ID=71682148

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/073652 WO2020151725A1 (fr) 2019-01-21 2020-01-21 Procédé et dispositif de prédiction d'emplacement

Country Status (2)

Country Link
CN (1) CN111461380A (fr)
WO (1) WO2020151725A1 (fr)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140143187A1 (en) * 2012-11-19 2014-05-22 Ha Yoon Song Method for probabilistically predicting location of object
CN104680250A (zh) * 2015-02-11 2015-06-03 北京邮电大学 一种位置预测系统
CN106202236A (zh) * 2016-06-28 2016-12-07 联想(北京)有限公司 一种用户位置预测方法及装置
CN106776930A (zh) * 2016-12-01 2017-05-31 合肥工业大学 一种融入时间和地理位置信息的地点推荐方法

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108072378B (zh) * 2016-11-15 2020-10-23 中国移动通信有限公司研究院 一种预测目的地的方法及装置
CN108288096B (zh) * 2017-01-10 2020-08-21 北京嘀嘀无限科技发展有限公司 用于估算行程时间、模型训练的方法及装置
CN107992530A (zh) * 2017-11-14 2018-05-04 北京三快在线科技有限公司 信息推荐方法及电子设备
CN108286980B (zh) * 2017-12-29 2021-04-02 广州通易科技有限公司 一种预测目的地和推荐驾驶路线的方法
CN108966148A (zh) * 2018-08-10 2018-12-07 深圳北斗应用技术研究院有限公司 出行信息的预测方法、服务器及存储介质

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140143187A1 (en) * 2012-11-19 2014-05-22 Ha Yoon Song Method for probabilistically predicting location of object
CN104680250A (zh) * 2015-02-11 2015-06-03 北京邮电大学 一种位置预测系统
CN106202236A (zh) * 2016-06-28 2016-12-07 联想(北京)有限公司 一种用户位置预测方法及装置
CN106776930A (zh) * 2016-12-01 2017-05-31 合肥工业大学 一种融入时间和地理位置信息的地点推荐方法

Also Published As

Publication number Publication date
CN111461380A (zh) 2020-07-28

Similar Documents

Publication Publication Date Title
TWI638320B (zh) 推薦預計到達時間的系統、方法及非暫態電腦可讀儲存媒體
TWI670677B (zh) 用於推薦預估到達時間的系統和方法
TWI673659B (zh) 用於確定預估到達時間的電子系統和方法及相關的非暫時性電腦可讀取媒體
TWI689744B (zh) 用於確定預估到達時間的系統和方法及相關的非暫時性機器可讀取儲存媒體
JP2019527871A (ja) 到着予定時刻を決定するシステム及び方法
US20210140774A1 (en) Systems and methods for recommending pick-up locations
CN112868036B (zh) 位置推荐的系统和方法
JP7047096B2 (ja) オンラインツーオフラインサービスに関する推定された到着時間を決定するためのシステムおよび方法
JP6503474B2 (ja) 移動デバイスの経路を求めるシステム及び方法
AU2018102202A4 (en) Systems and methods for cheat examination
CN109791731B (zh) 一种预估到达时间的方法和系统
WO2018227387A1 (fr) Procédés et systèmes de planification d'itinéraire
CN111415024A (zh) 一种到达时间预估方法以及预估装置
CN110751531A (zh) 轨迹识别方法、装置及电子设备
CN110800030B (zh) 用于拼车服务的方法和系统
CN111144968B (zh) 用于分配服务请求的系统和方法
CN111260423A (zh) 订单分配方法、装置、电子设备及计算机可读存储介质
WO2020151725A1 (fr) Procédé et dispositif de prédiction d'emplacement
CN113924460B (zh) 确定服务请求的推荐信息的系统和方法
CN111954891B (zh) 一种跨业务的共享汽车资源复用方法
CN114365205A (zh) 确定在线到离线服务中的预估到达时间的系统和方法
CN111954891A (zh) 一种跨业务的共享汽车资源复用方法
CN110832811A (zh) 用于发送空间数据的系统和方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20745052

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20745052

Country of ref document: EP

Kind code of ref document: A1