JP2001014297A - Method and device for predicting action and providing information - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an effective information corresponding to predicted action by predicting the action on the basis of unit action history information (ACTIV ITY). SOLUTION: A base station 21 transmits a position register signal, which is transmitted from portable terminal equipment 20 in a cycle corresponding to the moving speed, and ID information to an access server 31 of a communication industry device part 30. Every time the position register signal is transmitted from the portable terminal equipment 20, current position information expressing a position and time is stored in a position information storage part 35. Every time the new current position information of the portable terminal equipment 20 is stored in the position information storage part 35 or at every prescribed timing, the CPU of an action pattern analytic server 33 executes updating processing of an ACTIVITY object. At a request from a service provider 40, the action pattern analytic server 33 predicts the action of a user holding the portable terminal equipment 20 while using the ACTIVITY object stored in an action pattern information storage part 34.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a behavior prediction method, an information provision method, and an apparatus therefor. For example, an information provision method for predicting a behavior of a user having a terminal device and providing useful information to the user. It is suitable for application to the device.

[0002]

2. Description of the Related Art Conventionally, for example, PHS (Personal Handypho
ne System), the location registration signal transmitted from the PHS terminal device is received by the base station in the wireless zone where the PHS terminal is located, and the base station transmits the location registration signal to the PHS service control station from the base station. The station recognizes the position of the PHS terminal device for each wireless zone of the base station.

[0003] A system has been considered which uses the position information of the PHS terminal device to provide the PHS terminal device with various kinds of information according to the position of the PHS terminal device.

[0004]

However, in a system that provides information according to the current position of the terminal device, it is difficult to predict the behavior of the user who has the terminal device and provide the information according to the prediction. It is.

The present invention has been made in view of the above points, and provides a behavior prediction method, an information provision method, and a method for predicting a user's behavior and providing useful information according to the predicted behavior. It is intended to propose a device.

[0006]

SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the present invention detects a stop or a moving state of a prediction target, and performs an action of the prediction target at a cycle corresponding to the detected stop or a moving state of the prediction target. The history discrete information (EVENT) is sampled, and the sampled discrete information (EVE) is sampled.
NT) from the target action history information (ACTIVI)
TY), and the extracted unit action history information (ACT)
By predicting the behavior of the prediction target based on (IVITY), when the prediction target stops or moves less, the discrete information (EVENT) of the prediction target is sampled at a long sampling cycle, and the movement of the prediction target is performed. When the degree of the movement is large, the discrete information to be predicted (E
VENT) can be sampled.

Therefore, the behavior of the prediction target can be predicted with high accuracy by the discrete information (EVENT) sampled according to the stop or the moving state of the prediction target.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below in detail with reference to the drawings.

(1) Principle of Behavior Prediction Method The behavior prediction method according to the present invention uses a discrete system model composed of four elements. The four elements that make up this discrete system model are, first, E
NTITY and, second, an EVENT that represents a cross section of the occurring phenomenon as an event captured without considering time consumption
And third, an AC representing the operation or action performed by ENTITY.
TIVITY, and fourthly, an EVENT (or ACTIV) that focuses on a certain ENTITY and relates to the ENTITY.
PRO representing a phenomenon including a time lapse by a column of (ITY)
And CESS.

[0010] To show these elements in a concrete example, as shown in FIG. 1, for example, as an action of a customer who purchases a ticket at a ticket issuing machine, the customer is referred to as ENTITY which is a first element and a second element.
Arrival EVENT where the customer joins the queue of waiting customers at the ticketing machine as an EVENT, and the customer finishes the waiting queue and starts buying a ticket at the ticketing machine Start of service
EVENT and customers finish buying tickets at ticketing machines End of
service EVENT. The third factor, ACTIVITY, represents the act of a customer buying a ticket,
The fourth element, PROCESS, is EVE related to customers
This represents the NT column.

[0011] As described above, only the events (EVENT) having an important meaning with respect to the operation and action of the ENTITY (customer) are used, and the behavior of the customer is discretely modeled by the sequence (PROCESS). It is used for the behavior prediction.

Therefore, in this discrete system model, the time of an important event (EVENT) is counted, so that the time is unequally uneven depending on the event that occurs and the time when the event occurs.

Here, with respect to a certain subject (ENTITY), basic modeling can be performed by the subject's operation (ACTIVITY) and the processing period of the operation (the time when the activity starts and ends). That is, as shown in FIG.
A basic model is composed of a column of CTIVITY and an entity (ENTITY).

In this basic model, the action of the subject is regarded as an action of discrete change, and the time at which the action switches and the switch destination are expressed as a discrete model of the action pattern. The prediction is made.

Next, a method of discretely modeling an action pattern used for predicting the action of the subject will be described. In the method of generating this discrete system model, EN
The subject that is TITY is, for example, a user, the actual behavior (position and movement) of the user is detected by a predetermined position detection unit, and the stay and movement states of the user are described later using the detected results. M
ove Extract as ACTIVITY.

That is, as shown in FIG. 3, in the method of creating a discrete system model of an action pattern, the position of the user is sampled at a period determined according to the movement of the user, and the sampled position information and its time are sampled. To E
Store as VENT.

In this case, in the method of creating a discrete system model,
When the movement state of the user is detected by the movement detection means described later, the sampling cycle of the user position is changed according to the degree of the movement, for example, the moving speed of the user. That is, the sampling cycle changing means of the behavior prediction system that creates a discrete system model is configured to determine the sampling cycle according to the moving state of the user according to the sampling cycle changing procedure shown in FIG.

In FIG. 4, when the behavior prediction system enters the sampling cycle changing process from step SP30,
In the following step SP31, the movement detecting means detects a movement state comprising information such as the movement speed of the user, and notifies the detection result to a sampling cycle changing function unit (described later) in step SP32.

The sampling period changing function unit determines a sampling period of the user position based on the moving state, and samples the user position based on the determined sampling period. In this case, the sampling cycle changing function unit detects that the user has started moving when the detected moving speed of the user is equal to or higher than a predetermined speed.

In the behavior prediction system, in the sampling cycle change processing procedure, when the user is stopped at a fixed position, the sampling cycle of the user position by the position detecting means is made the longest, and the user moves Is detected, the sampling period is shortened continuously or stepwise as the moving speed increases.

Accordingly, in the user behavior pattern shown in FIG. 3, the sampling cycle of the user position (EVENT) is shortened in response to the user starting to move at the time T11, and the user is stopped at the time T12. Therefore, the sampling period of the user position (EVENT) becomes longer.

As described above, in the method of creating a discrete system model, the action prediction system changes the sampling period of the user position according to the moving speed of the user, so that the user is moving or the moving speed is reduced. More user positions are sampled in the high case and the user position when the user is moving (EVEN
T), that is, the moving route of the user is detected with higher accuracy. On the other hand, when the moving speed of the user is low or when the user is stopped, fewer user positions (EVENT) are sampled, and the load such as power consumption of the functional unit that executes the sampling operation is reduced.

Then, in the method of creating a discrete system model, the behavior prediction system extracts ACTIVITY based on the stored EVENT. In this case, of the plurality of stored EVENTs (FIG. 3), EVENT1 and EVENT2 are at the same position, and EVENT3 is
It is assumed that the position is a position immediately after the start of the movement with respect to ENT2, and that EVENT3 to EVENT8 are different from each other. It is assumed that EVENT8 to EVENT10 are at the same position.

According to the detection result of each EVENT,
Regarding EVENT1 and EVENT2, it can be seen that the user stays at the same place during this time zone. On the other hand, for EVENT3 to EVENT8, it can be seen that the user is moving during this time zone. Therefore, in the method of creating a discrete system model, EVENT3 is defined as ACTIV.
EVENT1 and EV
The ENT2 side is set to Stay ACTIVITY indicating stay,
Mov that represents the movement of the user from EVENT3 to EVENT8
e ACTIVITY. Also, EVENT8 to EV
ENT10 indicates that the user is staying at the same place during this time period, and thus EVE
NT8 is the time when ACTIVITY is replaced.
S represents the user's stay on the VENT8 to EVENT10 side
tay ACTIVITY. By the way, EVENT3 ~
The departure point in Move Activity of EVENT 8 is the stay point of Stay Activity before that, and the destination point of Move Activity is the relevant Move A point.
It is the stay point of Stay ACTIVITY following CTIVITY. As described above, the Move ACTIVITY indicates the start point, the destination point, and the required time (time points T11 to T1).
2).

In this manner, the EVENT column while the user is staying at the same position is collected into one Stay AC
TIVITY, and the EVENT column while the user is moving is collected into one Move ACTIVITY.
And

FIG. 5 shows a processing procedure for extracting the ACTIVITY from the EVENT. That is, in FIG. 5, when the behavior prediction system enters the processing procedure from step SP10, in a succeeding step SP11, it is determined whether or not a sampling period (access interval) for capturing EVENT, which is user position and time information, is 30 minutes or less. Judge. Here, EVENT captured with an access interval of 30 minutes or more has insufficient reliability in connection before and after generating ACTIVITY newly. Therefore, in this case, the behavior prediction system performs step SP
In step 11, a negative result is obtained, and the process proceeds to step SP12.

In step SP12, the behavior prediction system determines whether or not there is an existing Stay ACTIVITY corresponding to the EVENT acquired at this time, that is, an existing Stay ACTIVITY at the same point. If a negative result is obtained here, this means that the EVE captured at this time is
NT indicates that it does not exist in the existing Stay ACTIVITY, and that the access interval is 30 minutes or more. At this time, the behavior prediction system proceeds to step SP1.
Then, the process goes to 3 to store the EVENT data at this time in a temporary data storage area (Tmp Box).

On the other hand, if a positive result is obtained in step SP12, it means that the access interval of the EVENT taken at this time is 30 minutes or more, and the existing
This indicates that a similar EVENT exists in Stay ACTIVITY. At this time, the behavior prediction system proceeds to step SP14, and sets the EVENT taken in at this time as one of the populations of the corresponding existing ACTIVITY. The data storage area (ACTIVITY Box) storing the ACTIVITY data is updated.

As described above, an EVENT or an EVENT column having an access time of 30 minutes or more is taken into a population only when there is an existing Stay ACTIVITY having a similar population.

If a positive result is obtained in step SP11, this means that the access interval of the EVENT taken at this time is within 30 minutes. S
Moving to P15, the ACTENT or EVENT column fetched at this time is determined based on the correspondence relationship (FIG. 3) with the EVENT immediately before and immediately after the EVENT or EVENT column.
It is determined whether it is Y or Move ACTIVITY.

Incidentally, as a judgment in step SP15, the behavior prediction system determines that the data interval between the head and the end of the EVENT column is not less than 20 minutes and the same position (EVEN).
T), this action is regarded as a stay at a certain point (area). In addition, the behavior prediction system considers the distance and time interval between
Do not consider it as Move ACTIVITY.

Then, in step SP15, the Stay
When a determination result of ACTIVITY is obtained, the behavior prediction system proceeds to step SP16, and the existing Stay ACTIVIT made of a similar population (EVENT) is obtained.
It is determined whether or not Y exists. If an affirmative result is obtained here, this indicates that there is an existing Stay ACTIVITY similar to Stay ACTIVITY by the EVENT or EVENT sequence fetched at this time, and at this time, the action prediction system performs step SP16.
Then, in Step SP15 described above, Stay AC
A data storage area (Stay ACTIVITY Box) storing Stay ACTIVITY data is updated by regarding the EVENT or EVENT sequence determined as TIVITY as one of the populations of the existing ACTIVITY corresponding thereto. As a result, the number of EVENTs (the number of mothers) of the Stay ACTIVITY is increased, and the Stay ACTIVITY is increased.
The probability of occurrence of ITY will increase.

On the other hand, if a negative result is obtained in step SP16, this means that the Stay ACTIVI by the EVENT or the EVENT sequence fetched at this time.
This indicates that there is no existing Stay ACTIVITY similar to TY. At this time, the behavior prediction system proceeds to step SP17 and uses the EVENT or EVENT sequence determined to be Stay ACTIVITY in step SP15 described above. Stay ACTIVITY
Create

On the other hand, in the above-mentioned step SP15, the EVENT column captured at this time is Move A
When a determination result of CTIVITY is obtained, the behavior prediction system proceeds to step SP18, and moves to the existing Move ACTIVIT including a similar population (EVENT column).
It is determined whether or not Y exists. If a positive result is obtained here, this indicates that the existing Move A similar to the Move ACTIVITY by the EVENT column captured at this time.
This indicates that CIVITY exists, and at this time, the action prediction system proceeds to step SP20 and replaces the EVENT column determined as Move ACTIVITY in step SP15 with the corresponding existing ACTIVI.
A data storage area storing Move ACTIVITY data as one of the populations of TY (Move ACTIVITY
Box). Thereby, the Move ACTIV
By increasing the number of mothers at the waypoint of ITY (that is, EVENT), the probability of occurrence of the Move ACTIVITY increases.

On the other hand, if a negative result is obtained in step SP18, this means that there is no existing Move ACTIVITY similar to the Move ACTIVITY by the EVENT column fetched at this time, and at this time, the behavior prediction The system proceeds to step SP19
Then, in the above-mentioned step SP15, Move AC
Using the EVENT column determined to be TIVITY, a new
Create Move ACTIVITY.

By the way, in step SP15, Stay
If the result of the determination of ACTIVITY or Move ACTIVITY is not obtained, the behavior prediction system proceeds to step S
Moving to P13, the EVENT data at this time is stored in a temporary data storage area (Tmp Box).

Thus, according to the procedure shown in FIG. 5, when the interval for taking in EVENT is within 30 minutes, the behavior predicting system determines whether it is Stay ACTIVITY or Move A.
The determination of whether or not it is CTIVITY is started.

As described above, two types of ACTIVITY (S
As shown in FIG. 6, the user's behavior pattern divided into (tay ACTIVITY and Move ACTIVITY) is composed of a first tier behavior pattern model consisting of Stay ACTIVITY and a second tier behavior pattern model consisting of Move ACTIVITY. I do.

The action pattern model of the second layer is a Move ACTI having a moving direction from the starting point to the destination.
The first layer Stay ACTIVITY exists at the place where these Move ACTIVITY are replaced.

Each Stay ACTIVITY in the first layer and each Move ACTIVITY in the second layer constitute an object containing various information on the user's action. As shown in FIG. 7, each Stay ACTIVITY is information indicating a point, and a start time at which to start staying at the point (FIG. 3).
, The end time of the stay at the point (corresponding to the point T11 in FIG. 3), the key information (Key) such as the date, day of the week, and weather when the user's location information was sampled, and the Stay ACTIVITY. ACT before
Before ACTIVITY information that indicates IVITY (there is one or more ACTIVITY),
Represents ACTIVITY following ACTIVITY Next
ACTIVITY information (one or more of each ACTIVI
TY exists), number of mothers (EVENT number and ACT
IVITY), Stay ACTIV
The service used by the user in ITY and the user's preference information represented by the number of times, the Stay ACTI
It has information on the location of VITY (town information and the like) and ENTITY information including a user name and the like.

Here, as shown in FIG. 8, every time a new EVENT or EVENT sequence is generated, the action prediction system determines whether there is an existing ACTIVITY having the EVENT or EVENT sequence as a component as shown in FIG. The determination is made according to the procedure described above. As this criterion, a matter of whether or not the EVENT point is the same is used. If the same ACTIVITY exists, the behavior prediction system sets the newly generated EVENT
Alternatively, an EVENT sequence is added as a component of the existing ACTIVITY, and the number of the ACTIVITY matrix (EVEN
T number) (FIG. 7) is rewritten. In addition, the behavior prediction system determines the Before ACTIVITY of the ACTIVITY rewritten at this time in accordance with the EVENT generated at this time or the connection with the ACTIVITY before and after the EVENT column.
Rewrite the information or Next ACTIVITY information.

As shown in FIG. 9, each Move ACT
IVITY is information indicating a departure point, information indicating a destination point, a required time of the Move ACTIVITY (corresponding to time points T11 to T12 in FIG. 3), a date, a day of the week, and weather when user position information is sampled. Key information (Key), information on waypoints (EVENT) (the number of mothers at each point, and a plurality of patterns exist together with their occurrence probabilities), and the services used by the user in the Move Activity and the number of times. It has user preference information, information on the movement route of the Move ACTIVITY (town information and the like), and ENTITY information including a user name and the like.

As for this Move ACTIVITY,
As in the case of Stay ACTIVITY, each time a new EVENT or EVENT sequence is generated, the behavior prediction system determines whether there is an existing ACTIVITY having these EVENT or EVENT sequence as constituent elements. A condition for this determination is that the departure point and the destination point are the same. And the same ACTIVIT
If Y exists, the behavior prediction system replaces the newly generated EVENT or EVENT string with the existing ACTIV
In addition to the ITY components (via points),
The number of mothers (EVENT number) at the waypoint of IVITY (FIG. 9) is rewritten.

In this way, Stay ACTIVITY
The object and the Move Activity object are newly generated EVENT or EVENT, respectively.
The column updates information such as the number of mothers. This number of mothers is used as an occurrence probability of ACTIVITY in behavior prediction described later.

Next, the accumulated Stay ACTIVITY
A method of predicting a user's action using an object and a Move ACTIVITY object will be described.

Each of the accumulated Stay Activity objects and Move Activity objects is
An ACTIVITY object having key information (Key) such as a day of the week and weather when each ACTIVITY occurs (FIGS. 7 and 9), and having key information (Key) matching the day of the week and weather to be predicted Is a selection target object.

Then, the behavior prediction system retrieves an ACTIVITY object as a prediction candidate using information included in each ACTIVITY object, such as a time zone to be predicted and a departure point, as a retrieval key. For example,
When the user designates the time zone and the weather from 7:00 am to 5:00 pm on Sunday, the behavior prediction system determines from the stored ACTIVITY objects.
ACTIVI having the time zone and key information (Key)
Search for a TY object.

Then, the behavior prediction system calculates the plurality of retrieved ACTIVITY objects.
Multiple ACTIVIs based on the location information and context
Create a plurality of action patterns connecting TY objects. The context of the ACTIVITY object means that the Stay ACTIVITY object has its Before ACTIVITY information (FIG. 7) and Next A
Using CTIVITY information (Fig. 7), and using Move AC
In the TIVITY object, its start point and destination point (FIG. 9) are used.

For example, as shown in FIG. 10, when the user designates a sunny day from 7:00 in the morning to 5:00 in the evening as a behavior prediction and sets the start point to home, the behavior prediction system ACTIVITY object and Move A that have as key information (Key)
An active object in a specified time zone is searched as a candidate from among the active objects (FIG. 10A), and the searched candidate object A
From the group N1 of O1, AO2,..., AOn, for example, a Stay ACTIVITY object AO1 of “I am at home” is set as a start object (FIG. 10B).

Then, the behavior prediction system uses the following ACTIVE objects as Move ACTIVITY objects AO2 starting from the home,
Specified in the Next Activity information of the Stay Activity object AO1 saying "I am at home"
Predict Move AACTIVITY object AO3 etc. A
Select as CTIVITY object.

Thus, each ACTIV selected
An ACTIVITY object that can follow the ITY object is selected. In this case, the selected ACTI
There may be a plurality of VITY objects, so that a plurality of predicted behavior patterns (PROCESS) are created by connecting the ACTIVITY objects.

As described above, the behavior prediction system uses each ACT
Information indicating the context of the IVITY object (Before
By creating PROCESS using e-ACTIVITY information, Next-ACTIVITY information) and information indicating the connection (starting point, destination point), if it is an ACTIVITY object included in the time zone to be predicted,
Even if the occurrence times are different, the connection of the ACTIVITY objects having the connection characteristic of the user's action pattern can be obtained as a candidate for action prediction.

Incidentally, the behavior prediction system of this embodiment defines two rules as rules for forming a row of ACTIVITY objects. The first rule is that the Stay Activity object and the Stay A
Move A between the CTIVITY object
Assume that a CTIVITY object exists. Thereby, it is possible to avoid occurrence of an unnatural behavioral break. And as a second rule, Move ACTIVE
Stay ACTIVITY before and after the TY object
Assume that both the object and the Move ACTIVITY object have the potential to be selected. However,
Successive Move Activity objects must be S
Suppose that it must be surrounded by a tay ACTIVITY object.

As described above, the behavior prediction system shown in FIG.
0 (B), a plurality of prediction patterns (PROC
ESS) can be created. By the way, each ACTI
The VITY object has an occurrence probability depending on the number of EVENTs and the number of waypoints. The probability of occurrence is calculated from one ACTIVITY object to another ACTIVIT
Y is used as a probability when transitioning to the Y object. As a result, the probability of occurrence of each PROCESS
It is obtained by the product of the occurrence probabilities of VITY.

FIG. 11 shows a result of predicting an action pattern including such an occurrence probability. In FIG. 11, ACTIV
The probability of transition from ITYa to ACTIVITYi is AC
The probability of occurrence of TIVITYi is 0.6, and the probability of transition from ACTIVITYi to ACTIVITYf is the probability of occurrence of ACTIVITYf of 0.
55. Therefore, for example, ACTIVITYa-AC
TIVITYi-ACTIVITYf-ACTIVIT
The probability of occurrence of PROCESS consisting of Yc-ACTIVITYk is determined by each of the ACTIVIs that constitute the PROCESS.
It is 0.33 which is the product of the occurrence probabilities of TY.

As described above, the behavior prediction system uses the AC retrieved by the set time zone and key information (Key).
A plurality of PROCESSes can be created from the TIVITY group based on the user's action history.

(2) Configuration of Behavior Prediction System FIG. 12 shows an information providing system 1 using the behavior prediction system.
0, the position of a user who has a portable terminal device 20 such as a PHS as a terminal device is detected by the communication business device unit 30, and the behavior pattern analysis server 33 of the communication business device unit 30 detects the position based on the position information. The user's behavior pattern is analyzed.

That is, as shown in FIG. 13, the portable terminal device 20 carried by the user has a CP on the data bus BUS.
A transmission / reception circuit unit (RF) 20E for transmitting / receiving a signal to / from the U20A, the memory 20B, and the base station 21, and an RF (Radio Frequency) signal received by the transmission / reception circuit unit 20E is converted into a baseband signal and transmitted. An MMI (Man Machine Interface) which is an interface with a baseband processing unit 20D that converts a baseband signal into an RF signal, a microphone 20G, and a speaker 20H.
e) The unit 20F, the display unit 20I, and the keyboard 20J are connected.

The CPU 20A executes various operations in accordance with the operation programs stored in the memory 20B, and controls each circuit according to the operations. Various processing contents of the CPU 20A are displayed on a display unit 201 constituted by a liquid crystal display panel or the like as necessary.

When the user inputs a telephone number of a desired destination, the keyboard 20J supplies data representing the telephone number to the CPU 20A. The CPU 20A transmits a connection request to the communication destination represented by the telephone number input by the user via the transmission / reception circuit unit 20E. At this time, the communication line 22 connects the line according to the response of the called party. When the line is connected, the transmission / reception circuit unit 20E supplies the RF signal from the called party received via the antenna to the baseband processing unit 20D, where the RF signal is converted into a baseband signal. By supplying the converted baseband signal to the MMI unit 20F, the baseband processing unit 20D converts the received audio signal from the call destination into the speaker 2
Output as audio from 0H.

When the user inputs a voice through the microphone 20G, the MMI unit 20F turns on the microphone 2G.
The input audio signal supplied from 0G is supplied to the baseband processing unit 20D, where the baseband signal is converted into an RF signal. Then, the baseband processing unit 20D sends the converted RF signal to the communication line 22 via the transmission / reception circuit unit 20E, thereby transmitting the RF signal to a call destination connected to the line.

The CPU 20A transmits various information input by the user operating the keyboard 20J to the communication destination via the baseband processing unit 20D and the transmission / reception circuit unit 20E, and superimposes information from the communication destination. The obtained RF signal is captured via the transmission / reception circuit unit 20E and the baseband processing unit 20D, and displayed on the display unit 20I.

Thus, the user using the portable terminal device 20 can have a conversation with the other party and exchange various information.

Here, the portable terminal device 20 has a GPS (Glo
bal Positioning System) 20P is provided and G
The PS 20P specifies the position of the mobile terminal device 20 based on the correlation of signals transmitted from a plurality of satellites.

The CPU 20A of the portable terminal device 20
Calculates the amount of change (moving speed) of the position of the portable terminal device 20 per unit time based on the position information obtained by the GPS 20P, and supplies the calculated moving speed to the sampling cycle changing unit 20S.

The sampling period changing unit 20S is provided with a CPU
A position registration signal, which will be described later, is transmitted to the access server 31 of the communication business device unit 30 at a cycle corresponding to the moving speed calculated by the 20A. That is, as shown in FIG. 14, the sampling cycle changing unit 20S amplifies the oscillation output of the oscillator 121 in the amplifier circuit 122 and supplies the amplified output to a plurality of frequency divider circuits 123 that divide the frequency by different frequency division ratios. Each frequency dividing circuit 123 generates a different frequency signal by dividing the oscillation output by the frequency dividing ratio assigned to each, and supplies this to the selector 124.

Further, the table section 125, having received the moving speed information D21 from the CPU 20A (FIG. 13), generates selection data D22 corresponding to the moving speed represented by the moving speed information D21, and supplies this to the selector 124. I do.

Accordingly, the selector 124 selects the frequency signal D23 corresponding to the moving speed supplied from the CPU 20A, and supplies this to the CPU 20A (FIG. 13). CPU
20A transmits a position registration signal, which will be described later, to the access server 31 of the communication business device unit 30 at the cycle of the frequency signal D23 supplied from the sampling cycle changing unit 20S.

That is, the CPU 20A of the portable terminal device 20
The position registration signal and the identification information of the mobile terminal device 20 (ID information including a telephone number, etc.) are transmitted to the base station 21 in the wireless zone where the mobile terminal device 20 is located in a cycle corresponding to the moving speed of the mobile terminal device ) Has been made to send.

For example, when the portable terminal device 20 is stopped (ie, when the moving speed is the lowest), the CPU
20A transmits the position registration signal and the identification information to the base station 21 at the preset longest cycle, while detecting that the mobile terminal device 20 has reached the preset predetermined speed, The CPU 20A is the mobile terminal device 20
Is determined to be moving, and the transmission cycle of the position registration signal and the identification information is shortened as the moving speed increases.

The base station 21 transmits the position registration signal and the I signal transmitted from the portable terminal device 20 at a cycle corresponding to the moving speed.
The D information is transmitted to the access server 31 of the communication business device unit 30. Accordingly, the access server 31 can recognize the position of the mobile terminal device 20 in units of the wireless zone of the base station, and obtain the current position information of the mobile terminal device 20 obtained together with the time information by, for example, a plurality of hard disks. It is stored in the configured position information storage unit 35.

The current position information stored in the position information storage unit 35 is EVENT information (FIG. 3) in which the portable terminal device 20 is ENTITY (FIGS. 1 and 2). Therefore, each time a position registration signal is transmitted from the mobile terminal device 20, the position information storage unit 35 stores current position information indicating the position and time.

Here, the behavior pattern analysis server 3 of the behavior prediction system for creating an ACTIVITY object
Numeral 3 has a CPU and a memory connected to the data bus, and the CPU executes the procedure of creating the ACTIVE object shown in FIG. 5 according to a program stored in the memory. Then, the CPU stores the created ACTIVITY object in the action pattern information storage unit 34 including, for example, a plurality of hard disks.

Further, the CP of the behavior pattern analysis server 33
U stores new current position information of the portable terminal device 20 (ie, EVENT or EVENT column) in the position information storage unit 35.
The update processing of the ACTIVITY object described above with reference to FIG. 8 is executed every time is stored or at a predetermined timing. As a result of this update processing, the ACTIVITY object stored in the behavior pattern information storage unit 34 increases the number of its mothers so that
The occurrence probability of the ITY object approaches a value reflecting the behavior pattern of the user, and a more accurate behavior pattern (ACTIVITY) can be obtained.

In this manner, in a state where the user's action pattern is stored in the action pattern information storage section 34 as an ACTIVITY object, the action pattern analysis server 33 stores the action pattern information storage in response to a request from the service provider 40. AC stored in the unit 34
Using the TIVITY object, the behavior of the user having the portable terminal device 20 is predicted by the method described above with reference to FIGS.

The server 41 of the service provider 40
As shown in FIG. 15, C connected to the data bus BUS
PU 41A, memory 41B, communication interface 41
C and a database 41D, and the CPU 41A executes various processes in accordance with programs stored in the memory 41B.

That is, the CPU 41A receives provided information from various subscribed terminals (not shown) via the network connected by the communication interface 41C,
This is stored in the database 41D. Such provided information is particularly useful information for a user who is in a specific area or point, such as a movie theater screening guide or a transportation operation status. Accordingly, the CPU 41A of the server 41 determines the information of the user (that is, the portable terminal owned by the user) predicted to go to the specific area or the specific point based on the time and the weather condition for providing the provided information. The information of the behavior pattern (that is, PROCESS) is requested to the behavior pattern analysis server 33 of the communication business equipment unit 30.

The behavior pattern analysis server 33 responds to the request by storing the A stored in the behavior pattern information storage unit 34.
10 and 11 using a CTIVITY object
The day of the week and weather designated by the service provider 40 by the method described above are used as key information (Key).
The behavior of the user having the mobile terminal device 20 is predicted.

In the behavior prediction process, the behavior pattern analysis server 33 generates a predicted behavior pattern (PROCESS), which is a connection between the ACTIVITY objects. In this case, the behavior pattern analysis server 33 generates a plurality of predicted behavior patterns (PROCES) having different occurrence probabilities.
S) is generated.

Then, the behavior pattern analysis server 33 finds, in the behavior prediction result, a relatively high occurrence of a user who is predicted to go to a specific area or point under the day of the week and the weather condition designated by the server 41 of the service provider 40. An action pattern (PROCESS) consisting of a probability is supplied to the server 41 of the service provider 40 together with information for specifying the user, that is, ID information such as a telephone number of the portable terminal device 20 owned by the user.

Thus, the CPU 41A of the server 41
Based on the behavior pattern supplied from the communication business device unit 30 and the user information thereof (information such as a telephone number specifying the mobile terminal device 20), the provided information read from the database for the mobile terminal device 20 is transmitted to the network ( FIG.
Parabolic antenna 25, satellite 24 and radio tower 23 shown in FIG.
Mobile terminal device 2 via a network comprising a communication line 22 and a network comprising a communication line 22 and a base station 21).
Send to 0.

In this case, the server 41 transmits the predicted behavior pattern (PROCES) supplied from the communication business unit 30.
User preference information (services used and number of times) contained in each Stay ACTIVITY object that constitutes S)
Information that is likely to be particularly necessary in the user's predicted behavior based on the information about the user and the location, the user's preference information (services used and number of times) included in each Move ACTIVITY object, and information about the travel route. Read from the database prior to user action,
This is transmitted to the portable terminal device 20.

As a method for selecting information that is highly likely to be needed by the user, the server 41 uses the user's preference information (services used and number of times) included in each ACTIVITY object supplied from the communication business equipment unit 30. Among them, a service whose number of times of use is higher than a predetermined threshold value is selected, and each ACTI is used.
Based on the information on the user's movement route and points included in the VITY object, information that is particularly effective on the user's behavior prediction route is selected and transmitted to the mobile terminal device 20.

Thus, the user possessing the portable terminal device 20 can enjoy in advance the information necessary for each time zone and place according to his / her preference in the action of the user. .

FIG. 16 shows an example of the arrangement of each functional unit in the behavior prediction system according to this embodiment.

In the information providing system 10 using the behavior prediction system, the terminal device (portable terminal device 20 such as PHS), the communication business device unit 30, and the behavior pattern analysis server 33 communicate with each other. It is designed to use a code that can be decrypted only by itself.
As this encryption method, a secret key (common key) encryption method or a public key encryption method is used.

The secret key cryptosystem is a system in which the transmitting side and the receiving side transmit and receive encrypted data using the same key, and do not disclose the key used for encrypting the data.
Specifically, there are DES (Data Encryption Standard) or triple DES, and these secret key systems have an effect that the encryption / decryption processing is fast.

Further, the public key method is divided into a secret key and a public key.
This is a method of encrypting data using a type of key, specifically, a method such as RSA, RC2 or RC4. In this public key cryptosystem, components (for example, the portable terminal device 2
0, communication business equipment unit 30, behavior pattern analysis server 3
When 3) increases, there is an effect that key management becomes easy.

(3) Operation and Effect of Embodiment In the above configuration, the behavior prediction system determines whether the user is stopped or moving based on the position detection information of the GPS 20P provided in the portable terminal device 20. When the detection is made and the user is stopped, the sampling cycle of the user position for obtaining EVENT is made longest.

When the user is stopped or the user is moving at a low speed, as shown in FIG. Location information (EVENT1A
~ EVENT6A, EVENT18A ~ EVENT25
A) is substantially the same information. Therefore, in the behavior prediction system, as described above with reference to FIG. 3, when the user is stopped or moving at a low speed, the position of the user is sampled in a long cycle, so that less user position information (EVENT1 to EVENT2, EVENT8 to EVENT8 to EVE
NT10) makes it possible to keep the detection accuracy of the user behavior to a practically sufficient level.

On the other hand, when the moving speed of the user increases, as shown in FIG. 17B, if the position of the user is sampled in a long cycle, the user action based on the sampling results (EVENT4B to EVENT9B) is performed. Detection accuracy deteriorates. Therefore, in the behavior prediction system, as described above with reference to FIG. 3, by shortening the sampling cycle as the moving speed of the user increases, the user behavior is determined by the user position information (EVENT3 to EVENT7) detected in the short sampling cycle. Can be maintained at a sufficient level.

Thus, the behavior prediction system predicts the behavior of the user based on the position information (EVENT) of the user detected at the sampling period corresponding to the moving speed of the user, and thereby the prediction accuracy is sufficient for practical use. Can be kept to a degree.

According to the above configuration, the period for sampling the position of the user is changed in accordance with the moving speed of the user. Therefore, when the user stops or moves at a low moving speed, the user position is changed. , The power consumption of the sampling function unit can be reduced without deteriorating the user position detection accuracy.

Further, in the behavior prediction system, when the moving speed of the user increases, the position of the user can be detected with high accuracy by shortening the sampling cycle of the user position.

Further, by determining the sampling cycle of the user position based on the movement detection result using the position detecting means such as the GPS 20P, the base station connected to the same place may change when the user moves between buildings or at the same place. In this case, or when the user moves slightly, it is possible to prevent the sampling period from being changed carelessly.

Further, in the behavior prediction system, the behavior prediction accuracy of the user can be maintained with high accuracy regardless of the moving speed of the user, so that the traffic congestion information,
Provision of accident information or information such as the time required to reach the destination can be performed with higher accuracy.

(4) Other Embodiments In the above-described embodiment, the mobile terminal device 20 includes a user movement detecting means including the GPS 20P and the CPU 20A, and the movement detecting means changes the sampling cycle of the user position. Although the case where the sampling period changing means is provided has been described, the present invention is not limited to this. For example, the position of the user is specified based on the position information transmitted from the GPS 20P provided in the mobile terminal device 20, and the specified position is specified. Movement detecting means for detecting the movement of the user based on the position information may be provided in the service provider 40 or the communication business equipment unit 30 side, and various arrangement examples can be applied as the arrangement of the components.
FIG. 18 shows an example of the arrangement of the constituent elements and the effect of each arrangement example.

When the movement detecting means is provided in the service provider 40 or the communication business equipment unit 30, the service provider 40 or the communication business equipment unit 30 predicts by the behavior pattern analysis server 33 when the user starts moving. Based on the user's action to be performed, it is determined whether the predicted action is temporally and financially optimal, and action navigation information based on the determination result is provided to the user (portable terminal device 20). Can be provided.

In the above-described embodiment, the case where the sampling period of the user position is shortened as the moving speed of the user increases is described. However, the present invention is not limited to this. Each of the two states of the moving state may be set to one sampling period, and the number of sampling periods to be set may be variously applied.

Further, in the above-described embodiment, the case where the movement of the user is detected based on the moving speed obtained based on the position information of the user has been described. However, the present invention is not limited to this. The position coordinates at the time of performing may be managed separately for each block (region), and it may be detected that the block moves from one block to another block. In this case, as a movement detection method, a position registration signal and identification information of the mobile terminal device 20 are transmitted from the mobile terminal device 20 carried by the user to the base station 21 in the wireless zone where the mobile terminal device 20 is located in a short cycle. The access server 31 may constantly transmit the block in which the mobile terminal device 20 is present.

In the above-described embodiment, the case where the movement of the user is detected based on the moving speed obtained based on the position information of the user has been described. However, the present invention is not limited to this. When moving from a position in a certain case by a predetermined distance or more, it may be detected that the user has started moving.

In the above-described embodiment, the case where the movement of the user is detected based on the moving speed obtained based on the position information of the user has been described. However, the present invention is not limited to this, and the speed change (speed May be determined to have started moving when the differential value (differentiated value of the second value) becomes equal to or greater than a predetermined value.

In the above-described embodiment, the case where the movement of the user is detected based on the moving speed obtained based on the position information of the user has been described. However, the present invention is not limited to this. Then, the moving means used by the user may be determined based on the moving speed. For example, in the behavior prediction system, if the detected moving speed of the user is equal to or lower than a predetermined moving speed, it is determined that the user is moving on foot, and the moving speed of the user is in response to the predetermined speed. If so,
It may be determined that a train or the like is used. By making such a determination, the behavior prediction system
It is easier to predict the user's stopovers and transportation modes.

In the above-described embodiment, a case has been described in which a PHS is used as the portable terminal device 20, but the present invention is not limited to this, and for example, a PDA (Personal Digital
gitalAssistant), a car navigation system, a personal computer, a portable television, a portable radio, and other various portable terminal devices.

Further, in the above-described embodiment, a case has been described in which various types of information on the predicted action area of the user are provided based on the result of the action prediction. However, the present invention is not limited to this, and the result of the action prediction is not limited to this. Various information may be applied as information to be provided, such as comparing the current position and notifying the user when the comparison result is different.

Further, in the above-described embodiment, the case where the service provider connected to the network is used has been described. However, the present invention is not limited to this, and various other information providing means can be used.

In the above embodiment, the case where a circuit having a plurality of frequency dividing circuits 123 is used as sampling period changing section 20S as described above with reference to FIG. 14 has been described. However, the present invention is not limited to this. For example, CP
In U20A, a sampling period according to the moving speed is calculated using a predetermined function SP = aV + b (where SP is the sampling period, V is the moving speed, and a and b are constants). A location registration signal may be transmitted.

[0108]

As described above, according to the present invention, the stop or moving state of the prediction target is detected, and the discrete information of the action history of the prediction target is generated at a cycle corresponding to the detected stop or movement state of the prediction target. By sampling, extracting unit behavior history information of the prediction target from the sampled discrete information, and predicting the behavior of the prediction target based on the extracted unit behavior history information, according to the stop or moving state of the prediction target The behavior of the prediction target can be predicted with high accuracy by the discrete information sampled by the sampling.

[Brief description of the drawings]

FIG. 1 is a schematic diagram for explaining a discrete system model according to the present invention.

FIG. 2 is a schematic diagram used for describing a method of expressing a discrete system model.

FIG. 3 is a schematic diagram illustrating a method of extracting ACTIVITY from an EVENT column.

FIG. 4 is a flowchart showing a procedure for changing a sampling period of a user position;

FIG. 5 is a flowchart illustrating a method of extracting ACTIVITY from an EVENT column.

FIG. 6: Stay ACTIVITY and Move ACTIV
It is a basic diagram used for description of ITY.

FIG. 7 is a schematic diagram illustrating information assigned to a Stay ACTIVITY object.

FIG. 8 is a block diagram showing a method of updating ACTIVITY.

FIG. 9 is a schematic diagram illustrating information assigned to a Move ACTIVITY object.

FIG. 10 is a schematic diagram used for describing an action prediction process.

FIG. 11 is a schematic diagram illustrating a prediction result of a behavior pattern of a user.

FIG. 12 is a schematic connection diagram illustrating an overall configuration of an information providing system using the behavior prediction system according to the present invention.

FIG. 13 is a block diagram illustrating a configuration of a portable terminal device.

FIG. 14 is a configuration of a sampling cycle changing unit.

FIG. 15 is a block diagram illustrating a configuration of a server.

FIG. 16 is a block diagram illustrating an example of the arrangement of components.

FIG. 17 is a schematic diagram illustrating user position information sampled at different sampling periods.

FIG. 18 is a block diagram showing an example of the arrangement of components according to another embodiment.

[Explanation of symbols]

10 Information providing system, 20 Portable terminal device, 2
0A, 41A ... CPU, 20B, 41B ... memory
20S ... Sampling period change unit, 20P ... GP
S, 21 base station, 22 communication line, 30 communication business unit, 31 access server, 33 behavior pattern analysis server, 34 behavior pattern storage, 35
…… Location information storage unit, 40 …… Service provider, 4
1. Server.

 ──────────────────────────────────────────────────続 き Continued on the front page F-term (reference)

Claims (19)

[Claims] 1. A behavior predicting method for predicting a behavior of a prediction target, comprising: a movement detecting step of detecting a stop or a moving state of the prediction target; and a period according to the detected stop or a moving state of the prediction target. An action history obtaining step of sampling discrete information of the action history of the prediction target; a unit action history extraction step of extracting the unit action history information of the prediction target from the obtained discrete information; and the extracted unit action history A behavior prediction step of predicting the behavior of the prediction target based on the information. 2. The behavior predicting method according to claim 1, wherein the movement detecting step detects a moving state of the prediction target when a movement speed of the prediction target reaches a predetermined value. 3. The action history acquiring step, wherein the discrete information is sampled at a longest cycle when the prediction target is stopped, and when the prediction target is moving, the discrete information is sampled more than the sampling period in the stopped state. The behavior predicting method according to claim 1, wherein the discrete information is sampled in a short cycle. 4. The movement detecting step detects a degree of movement of the prediction object as a movement state of the prediction object, and the action history obtaining step includes performing the prediction at a cycle corresponding to the detected degree of movement. The behavior prediction method according to claim 1, wherein discrete information of the behavior history of the target is sampled. 5. The action history acquiring step, wherein the discrete information is sampled at a longest cycle when the prediction target is stopped, and when the degree of movement is larger when the prediction target is moving, the discrete information is larger. The behavior prediction method according to claim 4, wherein the sampling period is shortened. 6. The behavior predicting method according to claim 4, wherein the movement detecting step detects a moving speed of the prediction target as a degree of movement of the prediction target. 7. A behavior predicting apparatus for predicting a behavior of a prediction target, comprising: a movement detecting means for detecting a stop or a moving state of the prediction target; and a cycle corresponding to the detected stop or a moving state of the prediction target. Action history acquisition means for sampling discrete information of the action history of the prediction target; unit action history extraction means for extracting the unit action history information of the prediction target from the obtained discrete information; and the extracted unit action history A behavior prediction device, comprising: behavior prediction means for predicting the behavior of the prediction target based on information. 8. The behavior prediction device according to claim 7, wherein said movement detection means detects a movement state of said prediction target when a movement speed of said prediction target reaches a predetermined value. 9. The behavior history acquiring means samples the discrete information at a longest cycle when the prediction target is stopped, and performs a sampling at a longer cycle than the sampling period in the stopped state when the prediction target is moving. The behavior predicting apparatus according to claim 7, wherein the discrete information is sampled in a short cycle. 10. The movement detecting means detects the degree of movement of the prediction target as the movement state of the prediction target, and the action history acquiring means performs the prediction at a cycle corresponding to the detected degree of movement. The behavior prediction device according to claim 7, wherein discrete information of the behavior history of the target is sampled. 11. The behavior history obtaining means samples the discrete information at the longest cycle when the prediction target is stopped, and when the prediction target is moving, increases the degree of the discrete information when the prediction target is moving. The behavior prediction device according to claim 10, wherein the sampling period of (1) is shortened. 12. The behavior predicting apparatus according to claim 10, wherein said movement detecting means detects a moving speed of said prediction target as a degree of movement of said prediction target. 13. A behavior prediction device for predicting a behavior of a prediction target, comprising: behavior history acquisition means for sampling discrete information of the behavior history of the prediction target at a period according to a stop or a moving state of the prediction target; Unit behavior history extracting means for extracting the unit behavior history information of the prediction target from the obtained discrete information, and behavior prediction means for predicting the behavior of the prediction target based on the extracted unit behavior history information. An action prediction device characterized by the following. 14. The behavior history acquiring means samples the discrete information at a longest cycle when the prediction target is stopped, and performs a sampling at a longer time than the sampling period in the stopped state when the prediction target is moving. 14. The behavior prediction device according to claim 13, wherein the discrete information is sampled in a short cycle. 15. The movement detecting means detects the degree of movement of the prediction target as the movement state of the prediction target, and the action history obtaining means detects the degree of movement of the prediction target according to the degree of movement being the movement state of the prediction target. 14. The behavior prediction device according to claim 13, wherein discrete information of the behavior history of the prediction target is sampled in a cycle. 16. The action history acquiring means samples the discrete information at the longest cycle when the prediction target is stopped, and increases the discrete information as the degree of movement increases when the prediction target moves. The behavior predicting apparatus according to claim 15, wherein the sampling period of (1) is shortened. 17. An information providing method for providing information to an information providing target, comprising: a movement detecting step of detecting a stop or a moving state of the information providing target; and a response to the detected stop or a moving state of the information providing target. An action history acquisition step of sampling discrete information of the action history of the information provision target at a given cycle; a unit action history extraction step of extracting the unit action history information of the information provision target from the acquired discrete information; A behavior prediction step of predicting the behavior of the information provision target based on the unit behavior history information obtained, and providing the information related to the prediction behavior to the information based on the predicted behavior of the information provision target predicted Transmitting an information to an object. 18. An information providing apparatus for providing information to an information providing object, comprising: a movement detecting means for detecting a stop or a moving state of the information providing object; Action history acquisition means for sampling the discrete information of the action history of the information provision target at a given cycle; unit action history extraction means for extracting the unit action history information of the information provision target from the acquired discrete information; Action prediction means for predicting the action of the information providing target based on the unit action history information obtained, and providing the information related to the predicted action based on the predicted predicted action of the information providing target to the information An information providing apparatus, comprising: an information transmitting unit that transmits information to an object. 19. An information providing apparatus for providing information to an information providing target, comprising: an action history obtaining means for sampling discrete information of the action history of the information providing target at a period according to a stop or a moving state of the information providing target; A unit action history extracting means for extracting the unit action history information of the information provision target from the acquired discrete information, and an action prediction for predicting the action of the information provision target based on the extracted unit action history information Means for transmitting information related to the predicted behavior to the information provision target based on the predicted predicted behavior of the information provision target.