WO2023175883A1 - 情報処理装置、情報処理方法、およびプログラム - Google Patents
情報処理装置、情報処理方法、およびプログラム Download PDFInfo
- Publication number
- WO2023175883A1 WO2023175883A1 PCT/JP2022/012558 JP2022012558W WO2023175883A1 WO 2023175883 A1 WO2023175883 A1 WO 2023175883A1 JP 2022012558 W JP2022012558 W JP 2022012558W WO 2023175883 A1 WO2023175883 A1 WO 2023175883A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- user
- information
- location information
- feature vector
- online
- Prior art date
Links
- 230000010365 information processing Effects 0.000 title claims abstract description 65
- 238000003672 processing method Methods 0.000 title claims description 6
- 239000013598 vector Substances 0.000 claims abstract description 95
- 238000012549 training Methods 0.000 claims abstract description 11
- 230000006399 behavior Effects 0.000 claims description 145
- 238000000034 method Methods 0.000 claims description 32
- 230000008569 process Effects 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 12
- 230000009471 action Effects 0.000 abstract description 9
- 238000013145 classification model Methods 0.000 description 17
- 238000004891 communication Methods 0.000 description 15
- 238000005516 engineering process Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 239000004973 liquid crystal related substance Substances 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002250 progressing effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0204—Market segmentation
-
- 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
Definitions
- the present invention relates to an information processing device, an information processing method, and a program, and particularly relates to a technology that uses information on users' online behavior.
- Patent Document 1 discloses a technology in which an access point in a wireless LAN acquires location information of a terminal device from a terminal device and detects that a user of the terminal device has visited a predetermined store. There is.
- Patent Document 1 With the technology disclosed in Patent Document 1, it is possible to grasp the locations visited by the user of the terminal device from the location information of the terminal device. On the other hand, the utilization of information (online information) regarding online behavior (online behavior) by users through web services is progressing. If a relationship can be established between the online information and offline information such as information on places that the user can visit, the offline information can be estimated from the online information. This makes it possible to provide effective advertisements and services related to the offline information to the user.
- the present invention has been made in view of the above-mentioned problems, and an object of the present invention is to provide a technology for building a relationship between information on online behavior and information on offline behavior of users.
- one aspect of the information processing device is configured to input online information of a user having location information and output an online behavior feature vector of the user having the location information.
- the information processing device includes a generation unit that generates an offline behavior feature vector of the user having the location information from the location information, and inputs the offline behavior feature vector of the user having the location information generated by the generation unit.
- the apparatus further includes a third learning means for training a third learning model so as to output information representing offline behavior of the user having the location information.
- the information representing the offline behavior of the user may include information on places the user is expected to visit.
- the information processing device includes a first estimating unit that inputs online information of a user who does not have location information into the first learning model and estimates an online behavior feature vector of the user who does not have the location information.
- a second estimation means for inputting the online behavior feature vector of the user who does not have the location information into a second learning model to estimate the offline behavior feature vector of the user who does not have the location information;
- the third estimating means inputs an offline behavior feature vector of a user who does not have the information into a third learning model to estimate information about the offline behavior of the user who does not have the location information.
- the online information of the user may be information regarding the online behavior of the user through the use of a web service.
- one aspect of the information processing method includes a step of inputting online information of a user having location information and outputting an online behavior feature vector of the user having the location information. a first learning step of learning the first learning model, and a second learning step of inputting the online behavior feature vector of the user having the location information and outputting the offline behavior feature vector of the user having the location information. and a second learning step of learning the learning model.
- the information processing method includes a generation step of generating an offline behavior feature vector of the user having the location information from the location information, and an offline behavior feature vector of the user having the location information output from the second learning model. further comprising a third learning step of training a third learning model to input information representing the offline behavior of the user having the location information, and in the third learning step, the The third learning model may be trained using the offline behavior feature vector of the user having the position information generated in the generation step.
- one aspect of the program according to the present invention is an information processing program for causing a computer to perform information processing, the program causing the computer to transmit online information of a user having location information.
- a first learning process in which a first learning model is trained so as to input an online behavior feature vector of a user having the location information, and to output an online behavior feature vector of the user having the location information; and a second learning process for learning a second learning model so as to output the offline behavior feature vector of the user having the position information.
- the program further causes the computer to perform a generation process of generating an offline behavior feature vector of the user having the position information from the position information, and a generation process of generating an offline behavior feature vector of the user having the position information output from the second learning model.
- a third learning process for inputting an offline behavior feature vector and learning a third learning model to output information representing the offline behavior of the user having the location information;
- the third learning process may include a process of learning the third learning model using the offline behavior feature vector of the user having the position information generated in the generation process.
- FIG. 1 shows an example of the configuration of an information processing system.
- FIG. 2 shows an example of the functional configuration of the information processing device 10 according to the embodiment.
- FIG. 3 shows the procedure for generating an offline behavior feature vector.
- FIG. 4 shows an example of a schematic architecture of the online feature estimation model 111.
- FIG. 5 shows an example of a schematic architecture of the offline feature estimation model 112.
- FIG. 6 shows an example of a schematic architecture of the classification model 113.
- FIG. 7 is a schematic diagram for explaining a process for estimating offline behavior of a user without location information.
- FIG. 8 shows an example of the hardware configuration of the information processing device 10 and the user device 11.
- FIG. 9 shows a flowchart of processing executed by the information processing device 10.
- FIG. 1 shows a configuration example of an information processing system according to this embodiment.
- this information processing system includes an information processing device 10 and a plurality of user devices 11-1 to 11-N (N>1) used by arbitrary plurality of users 1 to N. ) and a user device 11-u used by user u.
- the user devices 11-1 to 11-N and 11-u may be collectively referred to as the user device 11 unless otherwise specified.
- the terms user device and user may be used interchangeably.
- the user device 11 is, for example, a device such as a smartphone or a tablet, and is capable of communicating with the information processing device 10 via a public network such as LTE (Long Term Evolution) or a wireless communication network such as a wireless LAN (Local Area Network). It is composed of The user device 11 has a display unit (display surface) such as a liquid crystal display, and each user can perform various operations using a GUI (Graphic User Interface) provided on the liquid crystal display. The operation includes various operations on content such as an image displayed on the screen, such as a tap operation, a slide operation, and a scroll operation using a finger, a stylus, or the like. Note that the user device 11 is not limited to the device shown in FIG.
- the user device 11 may be provided with a separate display screen.
- the user device 11 can log into a web service (Internet-related service) provided from the information processing device 10 or from another device (not shown) via the information processing device 10 and use the service.
- the web services may include Internet shopping malls, online supermarkets, or services related to communications, finance, real estate, sports, and travel provided via the Internet.
- the user device 11 can transmit information regarding the user of the user device 11 to the information processing device 10.
- the user device 11 can transmit information (online information) regarding online behavior (online behavior) by a user through a web service to the information processing device 10.
- the user device 11 performs positioning calculations based on signals received from GPS (Global Positioning System) satellites (not shown), and generates information obtained by the calculation as position information of the user device 11. , and can be transmitted to the information processing device 10.
- GPS Global Positioning System
- the information processing device 10 acquires online information and location information from the user device 11, and estimates an arbitrary user's offline behavior (for example, information on places that the user can visit) based on the information.
- the information processing device 10 is configured to be able to acquire the location information of the user devices 11-1 to 11-N (i.e., users 1 to N), and It is assumed that location information has not been acquired.
- users 1 to N may be referred to as users who have location information
- user u may be referred to as a user who does not have location information.
- the information processing device 10 first acquires online information and location information from the user devices 11-1 to 11-N.
- the information processing device 10 uses the online information and location information to learn various learning models for associating information on online behavior and information on offline behavior of the user.
- the information processing device 10 estimates the offline behavior of the user u of the user device 11-u through machine learning using the online information acquired from the user device 11-u and various learned models.
- the offline behavior is, for example, a place that the user is expected to visit (a place that can be visited next).
- an offline action may be any action associated with a location that the user is expected to visit.
- actions associated with a gas station (location) include going to a car-related store.
- FIG. 2 shows an example of the functional configuration of the information processing device 10 according to this embodiment.
- the information processing device 10 shown in FIG. 2 includes a location information acquisition unit 101, an online information acquisition unit 102, an online feature generation unit 103, an offline feature estimation unit 104, an online feature estimation unit 105, a classification unit 106, a learning unit 107, and a content creation unit.
- the learning model storage unit 108 includes a unit 108, an output unit 109, and a learning model storage unit 110.
- the learning model storage unit 110 stores an online feature estimation model 111, an offline feature estimation model 112, and a classification model 113. The various learning models will be described later.
- the location information acquisition unit 101 acquires the location information of the user of the user device from each of the user devices 11-1 to 11-N (users having location information).
- the position information includes a locus (track log) of the position of the user device over a certain period of time, that is, position history information.
- the location information can include location-related information regarding locations visited (stopped by) by the user.
- Place-related information includes information on (1) the sequence of visited places (a series of visited places), (2) the category of visited places, (3) the name of visited places, and (4) the type of visited places. It can be included.
- Each of the place-related information corresponds to a visited place sequence 33, a visited place category 34, a visited place name 35, and a visited place type 36 in FIG. 3, which will be described later.
- the location information acquisition unit 101 may be configured to generate the location-related information from the acquired location information.
- a sequence of visited places 33 indicates a sequence of positions of one or more places where the user device does not move for a certain period of time and the user is considered to have stayed for a certain period of time.
- the sequence of visited locations may be a sequence of point locations obtained from a trajectory of the user's locations.
- the visited place category 34 is a visited place category derived from the location of one or more places obtained from the visited place sequence 33 and given map information, and is, for example, a store, a school, This shows the basic classification for classifying the nature of places, such as hospitals and parks. Moreover, the store category may be further subdivided. For example, the visited place category 34 may indicate categories such as convenience stores, gas stations, and pet-related shops.
- the name of the visited place 35 indicates the name of the visited place derived from the location of one or more places obtained from the visited place sequence 33 and given map information. Note that if the name of the visited place cannot be acquired from map information or the like, the name 35 of the visited place may be no information.
- the visited place type 36 is textual information about one or more place types obtained by the visited place sequence 33.
- the type of visited place 36 can be derived, for example, from the visited place category 34 or the visited place name 35, and is a classification that is more detailed (at a lower level) than the visited place category 34. can be shown.
- the online information acquisition unit 102 acquires online information of the user of the user device from each of the user devices 11-1 to 11-N.
- online information indicates information regarding online actions (online actions) by users through web services.
- the online information may include (1) web service usage history, (2) general demographic data, (3) number of logins per month, and (4) information about the user's interests.
- Each of the online information corresponds to a web service usage history 42, demographic data 43, number of logins per month 44, and user interests 45 in FIGS. 4 and 7, which will be described later.
- the web service usage history 42 is information on the usage history of websites by users.
- the web service usage history 42 may include dates, purchase history and entry history in the web service, and viewing time of the web service site (corresponding to the display time on the user device 11).
- the demographic data 43 indicates demographic user attributes such as gender, age, area of residence, occupation, and family structure of the user registered for use of the web service.
- the number of logins per month 44 indicates the number of logins per month by the user to the web service.
- User interests 45 indicate information regarding the interests of users who are registered or newly registered to use the web service. The user's interests 45 may be information derived by the information processing device 10 or the user device 11 based on the web service usage history 42.
- the offline feature generation unit 103 generates an offline behavior feature vector from the location information acquired by the location information acquisition unit 101.
- the offline behavior feature vector is a feature vector representing the characteristics of the offline behavior described above. The offline behavior feature vector generation process will be described later using FIG. 3.
- the online feature estimation unit 104 estimates and generates an online behavior feature vector from the online information acquired by the online information acquisition unit 102.
- the online behavior feature vector is a feature vector representing the characteristics of the online behavior described above.
- the online feature estimation unit 104 estimates an online behavior feature vector using the learned online feature estimation model 111 stored in the learning model storage unit 110.
- the online feature estimation model 111 will be described later using FIG. 4.
- the offline feature estimation unit 105 estimates and obtains an offline behavior feature vector from the online behavior feature vector generated by the online feature estimation unit 104.
- the offline feature estimation unit 105 estimates an offline behavior feature vector using the learned offline feature estimation model 112 stored in the learning model storage unit 110.
- the offline feature estimation model 112 will be described later using FIG. 5.
- the classification unit 106 uses the offline behavior feature vector estimated by the offline feature estimation unit 105 to estimate and classify a label corresponding to the user's offline behavior. That is, the classification unit 106 estimates the user's offline behavior. In this embodiment, the classification unit 106 estimates a label using the learned classification model 113 stored in the learning model storage unit 110. The classification model 113 will be described later using FIG. 6.
- the learning unit 107 trains each of the online feature estimation model 111, offline feature estimation model 112, and classification model 113, and stores these trained learning models in the learning model storage unit 110.
- the content creation unit 108 Based on the estimated offline behavior of the user corresponding to the label estimated by the classification unit 106, the content creation unit 108 creates content suitable for the offline behavior.
- the content may be tangible content or may be intangible content such as digital content.
- the content creation unit 108 can create an advertisement suitable for the offline behavior.
- the output unit 109 outputs the label and offline behavior information estimated by the classification unit 106 and the content created by the content creation unit 108.
- the output may be any output process, and may be output to an external device via a communication I/F (communication I/F 87 in FIG. 8), or may be output to a display unit (display unit 86 in FIG. 8). may be displayed.
- the location information acquisition unit 101 and the online information acquisition unit 102 may be configured by the same module as an acquisition unit. Further, the online feature estimation section 104, the offline feature estimation section 105, and the classification section 106 may be configured as the same module as the estimation section.
- the learning unit 107 performs learning processing based on online information and location information acquired from the user devices 11-1 to 11-N.
- the offline feature generation unit 103 generates an offline behavior feature vector.
- Figure 3 shows the procedure for generating offline behavior feature vectors.
- the offline feature generation unit 103 first generates features included in the position information 31 of the user devices 11-1 to 11-N acquired by the position information acquisition unit 101 (or generated from the position information). Location related information 32 is acquired. Then, the offline feature generation unit 103 concatenates the sequence 33 of visited places, the category 34 of visited places, the name 35 of visited places, and the type 36 of visited places included in the place related information 32 ( Concatenation 37) and embedding on the feature vector space to generate an offline behavior feature vector 38.
- the offline behavior feature vector 38 is used as correct data when the offline feature estimation model 112 is trained. Further, the offline behavior feature vector 38 is used as input data when the classification model 113 is trained.
- FIG. 4 shows an example of a schematic architecture of the online feature estimation model 111.
- the online feature estimation model 111 is a learning model configured to input the user's online information 41, estimate and output the online behavior feature vector 46.
- the learning unit 107 inputs the user's online information 41 and trains the online feature estimation model 111 to output the online behavior feature vector 46.
- the online information 41 includes information on the web service usage history 42, demographic data 43, number of logins per month 44, and user interests 45.
- the online feature estimation model 111 is configured with multiple convolutional layers, performs processing equivalent to the encoder network (encoding part) in an autoencoder using a neural network on the online information 41, and processes the online behavior feature vector 46. Output.
- the learning unit 107 trains the online feature estimation model 111 using the online information 41 acquired from the user devices 11-1 to 11-N, and generates a trained online feature estimation model 111.
- the online behavior feature vector 46 is used as input data when the offline behavior estimation model 112 is trained.
- FIG. 5 shows an example of a schematic architecture of the offline feature estimation model 112.
- the offline feature estimation model 112 is a learning model configured to input an offline behavior feature vector and output an offline behavior feature vector.
- the learning unit 107 inputs the offline behavior feature vector and trains the offline feature estimation model 112 to output the offline behavior feature vector.
- the offline feature estimation model 112 may be configured with, for example, a neural network (encoder-decoder model) in which linear transformation is performed between layers.
- the learning unit 107 trains the offline feature estimation model 112 using the online behavior feature vector 46 as input data and the offline behavior feature vector 38 (FIG. 3) as correct data, and uses the learned offline A feature estimation model 112 is generated.
- the online behavior feature vector 46 as input data is a feature vector estimated using the online feature estimation model 111 from the online information 41 acquired from the user devices 11-1 to 11-N.
- the learning unit 107 causes the offline feature estimation model 112 to learn the relationship between the online behavior feature vector and the offline behavior feature vector for users whose location information is known.
- the learning procedure for the offline feature estimation model 112 is not limited to this, and the offline feature estimation model 112 may be trained using training data consisting of an online behavior feature vector and an offline behavior feature vector defined by other methods. good.
- FIG. 6 shows an example of a schematic architecture of the classification model 113.
- the classification model 113 is a classification model (classifier) that classifies a label corresponding to the offline behavior from the offline behavior feature vector (classifies the offline behavior).
- the learning unit 107 inputs the offline behavior feature vector and trains the classification model 113 to classify the offline behavior and output information representing the offline behavior.
- each label corresponds to a location that the user is expected to visit as an offline action.
- the learning unit 107 trains the classification model 113 using the offline behavior feature vector 38 as input data and the classification label 61 indicating the offline behavior as the correct label (correct data), and uses the learned classification model. 113 is generated.
- the classification label 61 as the correct label is indicated by the visited place category 33 and the visited place name 34 in the place-related information 32 that is included in the place-related information 32 used when generating the offline behavior feature vector 38.
- Location information In other words, information about the location is set as the next possible location to visit.
- the classification label 61 may be information about multiple locations or may be information about one location.
- FIG. 7 is a schematic diagram for explaining a process for estimating offline behavior of a user without location information.
- the information processing device 10 uses various learning models learned using location information and online information acquired from user devices 11-1 to 11-N (users having location information) to The offline behavior of user u (user without location information) is estimated based on online information obtained from device 11-u.
- Online information 71 from the user device 11-u is acquired by the online information acquisition unit 102.
- the online information 71 includes information about the user u's web service usage history 42, demographic data 43, number of logins per month 44, and user interests 45.
- the online feature estimation unit 104 estimates and generates an online behavior feature vector 72 by applying the online information 71 to the online feature estimation model 111.
- the offline feature estimation unit 105 estimates and obtains the offline behavior feature vector 73 by applying the online behavior feature vector 72 to the offline feature estimation model 112.
- the classification unit 106 After generating the offline behavior feature vector 73, the classification unit 106 applies the offline behavior feature vector 73 to the classification model 113 to estimate and classify a label (classification label 74) corresponding to the offline behavior of the user u. Thereby, the classification unit 106 can estimate the offline behavior of the user u, that is, the places that the user u is expected to visit.
- FIG. 8 is a block diagram showing an example of the hardware configuration of the information processing device 10 according to this embodiment.
- the information processing apparatus 10 according to this embodiment can be implemented on any single or multiple computers, mobile devices, or any other processing platform.
- FIG. 8 although an example is shown in which the information processing device 10 is implemented in a single computer, the information processing device 10 according to the present embodiment is implemented in a computer system including a plurality of computers. good. A plurality of computers may be connected to each other through a wired or wireless network so that they can communicate with each other.
- the information processing device 10 may include a CPU 81, a ROM 82, a RAM 83, an HDD 84, an input section 85, a display section 86, a communication I/F 87, and a system bus 88.
- the information processing device 10 may also include an external memory.
- a CPU (Central Processing Unit) 81 centrally controls operations in the information processing device 10, and controls each component (82 to 87) via a system bus 88, which is a data transmission path.
- a ROM (Read Only Memory) 82 is a nonvolatile memory that stores control programs and the like necessary for the CPU 81 to execute processing. Note that the program may be stored in a nonvolatile memory such as an HDD (Hard Disk Drive) 84 or an SSD (Solid State Drive), or an external memory such as a removable storage medium (not shown).
- a RAM (Random Access Memory) 83 is a volatile memory and functions as the main memory, work area, etc. of the CPU 81. That is, the CPU 81 loads necessary programs and the like from the ROM 82 into the RAM 83 when executing processing, and implements various functional operations by executing the programs and the like.
- the HDD 84 stores, for example, various data and information necessary when the CPU 81 performs processing using a program. Further, the HDD 84 stores various data, various information, etc. obtained by the CPU 81 performing processing using programs and the like.
- the input unit 85 includes a pointing device such as a keyboard and a mouse.
- the display section 86 is composed of a monitor such as a liquid crystal display (LCD). The display section 86 may function as a GUI (Graphical User Interface) by being configured in combination with the input section 85.
- GUI Graphic User Interface
- the communication I/F 87 is an interface that controls communication between the information processing device 10 and external devices.
- the communication I/F 87 provides an interface with a network and executes communication with an external device via the network. Via the communication I/F 87, various data, various parameters, etc. are transmitted and received with external devices.
- the communication I/F 87 may perform communication via a wired LAN (Local Area Network) or a dedicated line that complies with communication standards such as Ethernet (registered trademark).
- the network that can be used in this embodiment is not limited to this, and may be configured as a wireless network.
- This wireless network includes a wireless PAN (Personal Area Network) such as Bluetooth (registered trademark), ZigBee (registered trademark), and UWB (Ultra Wide Band). It also includes wireless LAN (Local Area Network) such as Wi-Fi (Wireless Fidelity) (registered trademark) and wireless MAN (Metropolitan Area Network) such as WiMAX (registered trademark). Furthermore, it includes wireless WAN (Wide Area Network) such as LTE/3G, 4G, and 5G. Note that the network only needs to connect each device so that they can communicate with each other, and the communication standard, scale, and configuration are not limited to the above.
- each element of the information processing device 10 shown in FIG. 8 can be realized by the CPU 81 executing a program. However, at least some of the functions of each element of the information processing device 10 shown in FIG. 8 may operate as dedicated hardware. In this case, the dedicated hardware operates under the control of the CPU 81.
- the hardware configuration of the user device 11 shown in FIG. 1 may be the same as that in FIG. 8. That is, the user device 10 may include a CPU 81 , a ROM 82 , a RAM 83 , an HDD 84 , an input section 85 , a display section 86 , a communication I/F 87 , and a system bus 88 .
- the user device 11 displays various information provided by the information processing device 10 on the display unit 86, and performs processing corresponding to input operations received from the user via the GUI (configured by the input unit 85 and the display unit 86). be able to.
- FIG. 9 shows a flowchart of processing executed by the information processing device 10 according to this embodiment.
- the process shown in FIG. 9 can be realized by the CPU 81 of the information processing device 10 loading a program stored in the ROM 82 or the like into the RAM 83 and executing it.
- FIG. 9 reference will be made to the information processing system shown in FIG. It is assumed that the online feature estimation model 111, offline feature estimation model 112, and classification model 113 that have been trained by the learning unit 107 are stored in the learning model storage unit 110 and used in the following processing.
- the online information acquisition unit 102 acquires the target user's online information (corresponding to the online information 71 in FIG. 7).
- the target user is user u (user device 11-u) shown in FIG.
- the online information 71 includes information on the web service usage history 42, demographic data 43, number of logins per month 44, and user interests 45.
- the online feature generation unit 103 applies the online information 71 acquired in S91 to the online feature estimation model 111 to generate the online behavior feature vector 72 of the user u.
- the offline feature estimation unit 105 applies the online behavior feature vector 71 of the user u generated in S92 to the offline feature estimation model 112 to generate the offline behavior feature vector 73 of the user u.
- the classification unit 106 applies the offline behavior feature vector 73 of the user u generated in S93 to the classification model 113 to generate a classification label 74. Further, the classification unit 106 estimates (specifies) the offline behavior of the user u that corresponds to the generated classification label 74.
- the content creation unit 108 creates content suitable for user u based on user u's offline behavior estimated in S94. For example, the content creation unit 108 creates an advertisement suitable for user u's offline behavior. If the location estimated as the offline behavior is a gas station, the content creation unit 108 may create an advertisement for a gas station or a car-related store or service.
- the output unit 109 outputs the information regarding the offline behavior of the user u estimated in S94 and/or the content created in S95.
- the output unit 109 transmits information regarding the offline behavior of the user u to an external device (not shown), and the external device can be used for marketing.
- the output unit 109 may also transmit the created content to other users who have similar characteristics (attributes) to the user u.
- the information processing device 10 utilizes online information acquired from a plurality of users to derive the relationship between offline behavior and online behavior.
- the information processing device 10 can then use the derived relationship to generate estimated offline behavior for a user who does not have offline information such as location information. This makes it possible to develop marketing using offline behavioral data.
Landscapes
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Description
上記した本発明の目的、態様及び効果並びに上記されなかった本発明の目的、態様及び効果は、当業者であれば添付図面及び請求の範囲の記載を参照することにより下記の発明を実施するための形態から理解できるであろう。
図1に、本実施形態による情報処理システムの構成例を示す。本情報処理システムは、その一例として、図1に示すように、情報処理装置10と、任意の複数のユーザ1~Nにより使用される複数のユーザ装置11-1~11-N(N>1)と、ユーザuにより使用されるユーザ装置11-uを含んで構成される。なお、以下の説明において、特に説明がない限り、ユーザ装置11-1~11-N、11-uをユーザ装置11と総称しうる。また、以下の説明において、ユーザ装置とユーザという語は同義に使用されうる。
なお、ユーザ装置11は、図1に示すような形態のデバイスに限らず、デスクトップ型のPC(Personal Computer)や、ノート型のPCといったデバイスであってもよい。その場合、各ユーザによる操作は、マウスやキーボードといった入力装置を用いて行われうる。また、ユーザ装置11は、表示面を別に備えてもよい。
また、ユーザ装置11は、GPS(Global Positioning System)衛星(不図示)から受信される信号等に基づいて測位計算を行い、当該計算により得られた情報を、ユーザ装置11の位置情報として生成し、情報処理装置10へ伝達することができる。
本実施形態による情報処理装置10は、まず、ユーザ装置11-1~11-Nからオンライン情報と位置情報を取得する。情報処理装置10は、当該オンライン情報と位置情報とを利用して、ユーザのオンライン行動の情報とオフライン行動の情報とを関係づけるための各種学習モデルを学習させる。そして、情報処理装置10は、ユーザ装置11-uから取得したオンライン情報と学習済みの各種学習モデルを用いた機械学習により、ユーザ装置11-uのユーザuのオフライン行動を推定する。オフライン行動は、例えば、ユーザが訪れると予想される場所(次に訪れうる場所)である。あるいは、オフライン行動は、ユーザが訪れると予測される場所に関連付けられたあらゆる行動であってもよい。例えば、ガソリンスタンド(場所)に関連付けられる行動として、自動車関連のお店へ行くことが含まれる。
図2に示す情報処理装置10は、位置情報取得部101、オンライン情報取得部102、オンライン特徴生成部103、オフライン特徴推定部104、オンライン特徴推定部105、分類部106、学習部107、コンテンツ作成部108、出力部109、および学習モデル記憶部110を備える。学習モデル記憶部110は、オンライン特徴推定モデル111、オフライン特徴推定モデル112、分類モデル113を記憶している。当該各種学習モデルについては後述する。
デモグラフィックデータ43は、ウェブサービスの利用のために登録されている、当該ユーザの性別、年齢、居住地域、職業、家族構成等の人口統計学的なユーザ属性を示す。
月当たりのログイン回数44は、ウェブサービスに対する、ユーザによる月当たりのログイン回数を示す。
ユーザの興味45は、ウェブサービスの利用のために登録されている、または、新たに登録されたユーザの興味に関する情報を示す。ユーザの興味45は、ウェブサービス利用履歴42に基づいて、情報処理装置10またはユーザ装置11により導出された情報であってもよい。
次に、本実施形態による学習部107による学習段階の処理(学習処理)について説明する。学習部107は、ユーザ装置11-1~11-Nから取得されたオンライン情報と位置情報に基づいて学習処理を行う。当該学習処理のためにまず、オフライン特徴生成部103が、オフライン行動特徴ベクトルを生成する。
図4に、オンライン特徴推定モデル111の概略アーキテクチャの例を示す。オンライン特徴推定モデル111は、ユーザのオンライン情報41を入力して、オンライン行動特徴ベクトル46を推定して出力するように構成される学習モデルである。学習部107は、ユーザのオンライン情報41を入力して、オンライン行動特徴ベクトル46を出力するように、オンライン特徴推定モデル111を学習させる。前述したように、本実施形態では、オンライン情報41は、ウェブサービス利用履歴42、デモグラフィックデータ43、月当たりのログイン回数44、ユーザの興味45の情報を含む。オンライン特徴推定モデル111は、複数の畳み込み層を有して構成され、オンライン情報41に対してニューラルネットワークを用いたオートエンコーダにおけるエンコーダネットワーク(エンコード部分)と同等の処理を行い、オンライン行動特徴ベクトル46を出力する。
オンライン行動特徴ベクトル46は、オフライン行動推定モデル112を学習させる際の入力データとして用いられる。
図5に、オフライン特徴推定モデル112の概略アーキテクチャの例を示す。オフライン特徴推定モデル112は、オフライン行動特徴ベクトルを入力して、オフライン行動特徴ベクトルを出力するように構成される学習モデルである。学習部107は、オフライン行動特徴ベクトルを入力して、オフライン行動特徴ベクトルを出力するように、オフライン特徴推定モデル112を学習させる。オフライン特徴推定モデル112は、例えば、層と層の間で線形変換(Linear transformation)が行われるニューラルネットワーク(エンコーダ・デコーダモデル)で構成されうる。
図6に、分類モデル113の概略アーキテクチャの例を示す。分類モデル113は、オフライン行動特徴ベクトルから、当該オフライン行動に対応するラベルを分類する(オフライン行動を分類する)分類モデル(分類器)である。学習部107は、オフライン行動特徴ベクトルを入力して、オフライン行動を分類してオフライン行動を表す情報を出力するように、分類モデル113を学習させる。本実施形態において、各ラベルは、オフライン行動として、ユーザが訪れると予測される場所に対応する。
続いて、位置情報のないユーザのオフライン行動を推定(推論)する処理について、図7を参照して説明する。図7は、位置情報のないユーザのオフライン行動の推定処理を説明するための模式図である。本実施形態では、情報処理装置10が、ユーザ装置11-1~11-N(位置情報を有するユーザ)から取得された位置情報とオンライン情報を用いて学習された各種学習モデルを用いて、ユーザ装置11-uから取得したオンライン情報に基づいてユーザu(位置情報を有さないユーザ)のオフライン行動を推定する。
オンライン行動特徴ベクトル72の生成後、オフライン特徴推定部105は、オンライン行動特徴ベクトル72をオフライン特徴推定モデル112に適用することにより、オフライン行動特徴ベクトル73を推定して取得する。
オフライン行動特徴ベクトル73の生成後、分類部106は、オフライン行動特徴ベクトル73を分類モデル113に適用して、ユーザuのオフライン行動に対応するラベル(分類ラベル74)を推定して分類する。これにより、分類部106は、ユーザuのオフライン行動、すなわち、ユーザuが訪れると予測される場所を推定することができる。
図8は、本実施形態による情報処理装置10のハードウェア構成の一例を示すブロック図である。
本実施形態による情報処理装置10は、単一または複数の、あらゆるコンピュータ、モバイルデバイス、または他のいかなる処理プラットフォーム上にも実装することができる。
図8を参照して、情報処理装置10は、単一のコンピュータに実装される例が示されているが、本実施形態による情報処理装置10は、複数のコンピュータを含むコンピュータシステムに実装されてよい。複数のコンピュータは、有線または無線のネットワークにより相互通信可能に接続されてよい。
CPU(Central Processing Unit)81は、情報処理装置10における動作を統括的に制御するものであり、データ伝送路であるシステムバス88を介して、各構成部(82~87)を制御する。
RAM(Random Access Memory)83は、揮発性メモリであり、CPU81の主メモリ、ワークエリア等として機能する。すなわち、CPU81は、処理の実行に際してROM82から必要なプログラム等をRAM83にロードし、当該プログラム等を実行することで各種の機能動作を実現する。
入力部85は、キーボードやマウス等のポインティングデバイスにより構成される。
表示部86は、液晶ディスプレイ(LCD)等のモニターにより構成される。表示部86は、入力部85と組み合わせて構成されることにより、GUI(Graphical User Interface)として機能してもよい。
通信I/F87は、ネットワークとのインタフェースを提供し、ネットワークを介して、外部装置との通信を実行する。通信I/F87を介して、外部装置との間で各種データや各種パラメータ等が送受信される。本実施形態では、通信I/F87は、イーサネット(登録商標)等の通信規格に準拠する有線LAN(Local Area Network)や専用線を介した通信を実行してよい。ただし、本実施形態で利用可能なネットワークはこれに限定されず、無線ネットワークで構成されてもよい。この無線ネットワークは、Bluetooth(登録商標)、ZigBee(登録商標)、UWB(Ultra Wide Band)等の無線PAN(Personal Area Network)を含む。また、Wi-Fi(Wireless Fidelity)(登録商標)等の無線LAN(Local Area Network)や、WiMAX(登録商標)等の無線MAN(Metropolitan Area Network)を含む。さらに、LTE/3G、4G、5G等の無線WAN(Wide Area Network)を含む。なお、ネットワークは、各機器を相互に通信可能に接続し、通信が可能であればよく、通信の規格、規模、構成は上記に限定されない。
図1に示すユーザ装置11のハードウェア構成は、図8と同様でありうる。すなわち、ユーザ装置10は、CPU81と、ROM82と、RAM83と、HDD84と、入力部85と、表示部86と、通信I/F87と、システムバス88とを備えうる。ユーザ装置11は、情報処理装置10により提供された各種情報を、表示部86に表示し、GUI(入力部85と表示部86による構成)を介してユーザから受け付ける入力操作に対応する処理を行うことができる。
図9に、本実施形態による情報処理装置10により実行される処理のフローチャートを示す。図9に示す処理は、情報処理装置10のCPU81がROM82等に格納されたプログラムをRAM83にロードして実行することによって実現されうる。図9の説明のために、図1に示した情報処理システムを参照する。学習部107により学習済みの、オンライン特徴推定モデル111、オフライン特徴推定モデル112、および分類モデル113は、学習モデル記憶部110に格納され、以下の処理で用いられるものとする。
Claims (9)
- 位置情報を有するユーザのオンライン情報を入力して、前記位置情報を有するユーザのオンライン行動特徴ベクトルを出力するように、第1の学習モデルを学習させる第1の学習手段と、
前記位置情報を有するユーザのオンライン行動特徴ベクトルを入力して、前記位置情報を有するユーザのオフライン行動特徴ベクトルを出力するように、第2の学習モデルを学習させる第2の学習手段と、
を有することを特徴とする情報処理装置。 - 前記位置情報から、前記位置情報を有するユーザのオフライン行動特徴ベクトルを生成する生成手段と、
前記生成手段により生成された前記位置情報を有するユーザのオフライン行動特徴ベクトルを入力して、前記位置情報を有するユーザのオフライン行動を表す情報を出力するように、第3の学習モデルを学習させる第3の学習手段を更に有することを特徴とする請求項1に記載の情報処理装置。 - 前記ユーザのオフライン行動を表す情報は、前記ユーザが訪れると予測される場所の情報を含むことを特徴とする請求項2に記載の情報処理装置。
- 位置情報を有さないユーザのオンライン情報を前記第1の学習モデルに入力して、前記位置情報を有さないユーザのオンライン行動特徴ベクトルを推定する第1の推定手段と、
前記位置情報を有さないユーザのオンライン行動特徴ベクトルを前記第2の学習モデルに入力して、前記位置情報を有さないユーザのオフライン行動特徴ベクトルを推定する第2の推定手段と、
前記位置情報を有さないユーザのオフライン行動特徴ベクトルを前記第3の学習モデルに入力して、前記位置情報を有さないユーザのオフライン行動の情報を推定する第3の推定手段と、
を有することを特徴とする請求項2または3に記載の情報処理装置。 - 前記ユーザのオンライン情報は、前記ユーザにウェブサービスの利用を通したオンライン上の行動に関する情報であることを特徴とする請求項1から4のいずれか1項に記載の情報処理装置。
- 位置情報を有するユーザのオンライン情報を入力して、前記位置情報を有するユーザのオンライン行動特徴ベクトルを出力するように、第1の学習モデルを学習させる第1の学習工程と、
前記位置情報を有するユーザのオンライン行動特徴ベクトルを入力して、前記位置情報を有するユーザのオフライン行動特徴ベクトルを出力するように、第2の学習モデルを学習させる第2の学習工程と、
を有することを特徴とする情報処理方法。 - 前記位置情報から、前記位置情報を有するユーザのオフライン行動特徴ベクトルを生成する生成工程と、
前記第2の学習モデルから出力された前記位置情報を有するユーザのオフライン行動特徴ベクトルを入力して、前記位置情報を有するユーザのオフライン行動を表す情報を出力するように、第3の学習モデルを学習させる第3の学習工程を更に有し、
前記第3の学習工程では、前記生成工程において生成された前記位置情報を有するユーザのオフライン行動特徴ベクトルを用いて、前記第3の学習モデルを学習させることを特徴とする請求項6に記載の情報処理方法。 - 情報処理をコンピュータに実行させるための情報処理プログラムであって、該プログラムは、前記コンピュータに、
位置情報を有するユーザのオンライン情報を入力して、前記位置情報を有するユーザのオンライン行動特徴ベクトルを出力するように、第1の学習モデルを学習させる第1の学習処理と、
前記位置情報を有するユーザのオンライン行動特徴ベクトルを入力して、前記位置情報を有するユーザのオフライン行動特徴ベクトルを出力するように、第2の学習モデルを学習させる第2の学習処理と、を含む処理を実行させるためのものである、
情報処理プログラム。 - 前記プログラムは、前記コンピュータに、更に、
前記位置情報から、前記位置情報を有するユーザのオフライン行動特徴ベクトルを生成する生成処理と、
前記第2の学習モデルから出力された前記位置情報を有するユーザのオフライン行動特徴ベクトルを入力して、前記位置情報を有するユーザのオフライン行動を表す情報を出力するように、第3の学習モデルを学習させる第3の学習処理と、を含む処理を実行させるためのものであり、
前記第3の学習処理は、前記生成処理において生成された前記位置情報を有するユーザのオフライン行動特徴ベクトルを用いて、前記第3の学習モデルを学習させる処理を含む、情報処理プログラム。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2022/012558 WO2023175883A1 (ja) | 2022-03-18 | 2022-03-18 | 情報処理装置、情報処理方法、およびプログラム |
EP22932156.7A EP4307202A1 (en) | 2022-03-18 | 2022-03-18 | Information processing device, information processing method, and program |
JP2023515763A JPWO2023175883A1 (ja) | 2022-03-18 | 2022-03-18 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2022/012558 WO2023175883A1 (ja) | 2022-03-18 | 2022-03-18 | 情報処理装置、情報処理方法、およびプログラム |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2023175883A1 true WO2023175883A1 (ja) | 2023-09-21 |
Family
ID=88022926
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2022/012558 WO2023175883A1 (ja) | 2022-03-18 | 2022-03-18 | 情報処理装置、情報処理方法、およびプログラム |
Country Status (3)
Country | Link |
---|---|
EP (1) | EP4307202A1 (ja) |
JP (1) | JPWO2023175883A1 (ja) |
WO (1) | WO2023175883A1 (ja) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2015037244A (ja) | 2013-08-13 | 2015-02-23 | ソフトバンクモバイル株式会社 | 情報照合装置、情報処理システム、プログラム及び照合方法 |
JP2019215862A (ja) * | 2018-06-07 | 2019-12-19 | 技研商事インターナショナル株式会社 | 行動履歴データを用いた行動分析システム及びそのプログラム |
JP2020161038A (ja) * | 2019-03-28 | 2020-10-01 | 株式会社博報堂Dyホールディングス | 推定システム、推定方法、及びコンピュータプログラム |
JP2020190964A (ja) * | 2019-05-22 | 2020-11-26 | ヤフー株式会社 | 情報処理装置、情報処理方法および情報処理プログラム |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20090024546A1 (en) * | 2007-06-23 | 2009-01-22 | Motivepath, Inc. | System, method and apparatus for predictive modeling of spatially distributed data for location based commercial services |
JP6899805B2 (ja) * | 2018-08-27 | 2021-07-07 | 楽天グループ株式会社 | 特性推定装置、特性推定方法、及び特性推定プログラム等 |
-
2022
- 2022-03-18 WO PCT/JP2022/012558 patent/WO2023175883A1/ja active Application Filing
- 2022-03-18 JP JP2023515763A patent/JPWO2023175883A1/ja active Pending
- 2022-03-18 EP EP22932156.7A patent/EP4307202A1/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2015037244A (ja) | 2013-08-13 | 2015-02-23 | ソフトバンクモバイル株式会社 | 情報照合装置、情報処理システム、プログラム及び照合方法 |
JP2019215862A (ja) * | 2018-06-07 | 2019-12-19 | 技研商事インターナショナル株式会社 | 行動履歴データを用いた行動分析システム及びそのプログラム |
JP2020161038A (ja) * | 2019-03-28 | 2020-10-01 | 株式会社博報堂Dyホールディングス | 推定システム、推定方法、及びコンピュータプログラム |
JP2020190964A (ja) * | 2019-05-22 | 2020-11-26 | ヤフー株式会社 | 情報処理装置、情報処理方法および情報処理プログラム |
Non-Patent Citations (1)
Title |
---|
See also references of EP4307202A4 |
Also Published As
Publication number | Publication date |
---|---|
JPWO2023175883A1 (ja) | 2023-09-21 |
EP4307202A4 (en) | 2024-01-17 |
EP4307202A1 (en) | 2024-01-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Workman | New media and the changing face of information technology use: The importance of task pursuit, social influence, and experience | |
Gheisari et al. | Investigating human and technological requirements for successful implementation of a BIM-based mobile augmented reality environment in facility management practices | |
Pinochet et al. | The influence of the attributes of “Internet of Things” products on functional and emotional experiences of purchase intention | |
JP2018190396A (ja) | ネットワークレーティング予測エンジン | |
US20180158163A1 (en) | Inferring appropriate courses for recommendation based on member characteristics | |
US11803872B2 (en) | Creating meta-descriptors of marketing messages to facilitate in delivery performance analysis, delivery performance prediction and offer selection | |
JP6986978B2 (ja) | 情報処理装置、情報処理方法、及び情報処理プログラム | |
Valls-Donderis et al. | Participatory development of decision support systems: which features of the process lead to improved uptake and better outcomes? | |
US10229513B2 (en) | Cognitive color scheme advisor for dynamic modification of digital media in a collaborative environment with color blind individuals | |
Ballatore et al. | Towards a usability scale for participatory GIS | |
Kumar | The moderating factors of 3G user acceptance technology in Shimla (India) using UTAUT model | |
Tkacz et al. | Data diaries: A situated approach to the study of data | |
Sia et al. | Smart mobile tourism app featuring augmented reality and big data analytics: an empirical analysis using UTAUT2 and PCT models | |
JP2021119526A (ja) | 生成装置、生成方法、及び生成プログラム | |
WO2023175883A1 (ja) | 情報処理装置、情報処理方法、およびプログラム | |
Al-Nuaim | Context-aware mobile interface design for M-government | |
US11163964B2 (en) | Configurable conversational agent generator | |
Tarka et al. | On the Unstructured Big Data Analytical Methods in Firms: Conceptual Model, Measurement, and Perception | |
US20210326392A1 (en) | Algorithmic attribution | |
Goktas et al. | Optimizing Customer Experience in Hospitality and Tourism with ChatGPT Plugins: A Strategic Guide | |
Lytras et al. | Policy implications for smart healthcare: the international collaboration dimension | |
US11907508B1 (en) | Content analytics as part of content creation | |
Reza et al. | Towards Designing Intuitive Mobile UIs Considering Tapping Behaviour of Elderly Users | |
JP7514878B2 (ja) | 情報処理装置、情報処理方法、およびプログラム | |
EP4254295A1 (en) | Information processing apparatus, information processing method, and program thereof |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 2023515763 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 22932156.7 Country of ref document: EP Ref document number: 2022932156 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref document number: 2022932156 Country of ref document: EP Effective date: 20231010 |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22932156 Country of ref document: EP Kind code of ref document: A1 |