JP2006134080A - Mobile terminal and personal adaptive context acquisition method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a personal digital assistant capable of acquiring a personal adaptive context capable of grasping individual behavior in detail and accurately, and a personal adaptive context acquisition method.
A mobile terminal 10 divides sensor data acquired by a sensor 15 for each event, and calculates a user specific function for approximating and expressing the divided sensor data. The mobile terminal 10 obtains the characteristic parameters e _all and C _all by developing the divided sensor data using the calculated user specific function. The mobile terminal 10 generates an SOM from the feature parameters e _all and C _all , and learns a correspondence relationship between the sensor information feature data mα, which is the output of the SOM, and the action data using a neural network model.
[Selection] Figure 4

Description

  The present invention relates to a portable terminal that acquires a personal adaptive context and a personal adaptive context acquisition method.

It is important to grasp the individual situation (context) in order to realize a real-world oriented service. As a conventional technique for grasping an individual context, a technique for acquiring a context based on a user's biometric information acquired by a sensor is known (for example, Non-Patent Document 1, Patent Document 1, And Patent Document 2).
According to the technique described in Non-Patent Document 1, sensor data representing the user's biological information and environmental information is acquired from the sensor by attaching an arm band on which the sensor is mounted to the user's arm. Each time the context (“walking”, “working”, “sleeping”, etc.) changes, the user presses a button on the armband to input information. The armband transmits the acquired sensor data to the computer, and the computer learns the correspondence between the sensor data and the context.

Further, in Patent Document 1, biometric information, environmental information, and action records of a user who owns a mobile terminal are stored in association with each other, and information on the user's actions based on these information (to the living body of the past schedule). The contents of the technique for outputting the analysis result of the influence and the prediction result of the influence of the future schedule on the living body are described.
Patent Document 2 describes the technical contents for estimating the user's behavior based on the user's body temperature and the temperature of the outside air.
JP-A-10-305016 (paragraphs 0058 to 62). Japanese Patent Laying-Open No. 2004-008471 (paragraph 0022) A. Krause et. Al, CyberCode: "Unsupervised, Dynamic Identification of Physiological and Activity Context in Wearable Computing", [searched on September 3, 2004], Internet <URL: http://www.cc.gatech.edu/ ccg / paper_of_week / krause-iswc2003-unsup-activity.pdf>

The conventional research on personal context acquisition has the following problems.
(1) Since the sensor data processing method is the same regardless of the user, it is not possible to extract individual-specific features from sensor data such as biometric information, and an accurate context for each individual cannot be obtained.
(2) Most of the contexts associated with the sensor data represent simple situations such as “walking”, “working”, “sleeping”, “being outdoors” It is not sufficient for use as information for intelligent service control, and the resulting context needs to be further interpreted.
The present invention has been made in view of the above points, and provides a mobile terminal and a personal adaptive context acquisition method for acquiring a personal adaptive context capable of grasping individual behavior in detail and accurately. For the purpose.

  In order to solve the above-mentioned problem, the invention described in claim 1 is a sensor data dividing unit that divides sensor data indicating information related to a user acquired by a sensor, and sensor data divided by the sensor data dividing unit. A user-specific function calculating unit for calculating a user-specific function for expressing the sensor, and a sensor for developing the sensor data divided by the sensor data dividing unit using the user-specific function calculated by the user-specific function calculating unit There is provided a portable terminal comprising: a data expansion unit; and a personal adaptive context learning unit that learns a correspondence relationship between a characteristic parameter obtained by expansion by the sensor data expansion unit and user behavior data. .

  According to this configuration, the mobile terminal learns the correspondence between the feature parameter obtained by developing the sensor data with the user-specific function and the behavior data of the user, and thus the unique feature specific to the individual user. Can learn. For this reason, based on this learning result, it is possible to obtain a personal adaptive context capable of grasping individual behavior in detail and accurately.

According to a second aspect of the present invention, in the mobile terminal according to the first aspect, the mobile terminal further comprises user behavior data storage means for storing service usage history in the terminal as user behavior data, and the personal adaptive context learning means is The correspondence relationship between the user behavior data stored in the user behavior data storage means and the feature parameter is learned.
According to this configuration, just by using the service using the mobile terminal, the mobile terminal can store the user's behavior data and learn the correspondence between the sensor data and the user's behavior data. Since it is not necessary to input data manually, it is possible to reliably learn correspondence.

According to a third aspect of the present invention, in the portable terminal according to the first or second aspect, the sensor data expansion means is configured to determine a feature parameter based on the user specific function and a difference between the sensor data and the user specific function. It is characterized by expanding to feature parameters based on.
According to this configuration, since the mobile terminal develops the feature parameters based on the user specific function and the feature parameters based on the difference between the sensor data and the user specific function, many user specific feature parameters are extracted from the sensor data. It is possible to accurately associate the characteristic parameter with the user's action data. Further, based on the degree of difference, it is possible to interpret whether or not the user's usual behavior pattern is taken.

The invention according to claim 4 corresponds to sensor data acquired by the sensor based on a learning result by the personal adaptive context learning means in the mobile terminal according to any one of claims 1 to 3. A personal adaptive context acquiring means for acquiring a user's action pattern as a personal adaptive context, and receiving a service corresponding to the personal adaptive context acquired by the personal adaptive context acquiring means. And
According to this configuration, since the mobile terminal acquires the personal adaptive context based on the result of learning the user-specific features, the personal adaptive context that can grasp the user's individual behavior in detail and accurately is provided. Can be earned. Then, the user can receive a service suitable for his / her context.

The invention according to claim 5 is a sensor data division step for dividing sensor data indicating information related to a user acquired by a sensor, and a user specific function for expressing the sensor data divided in the sensor data division step. Obtained by the user-specific function calculating step for calculating the sensor data, the sensor data expanding step for expanding the sensor data using the user specific function calculated in the user-specific function calculating step, and the expansion in the sensor data expanding step A personal adaptive context learning step for learning the correspondence between the feature parameter and the user behavior data, and corresponding to the sensor data acquired by the sensor based on the learning result in the personal adaptive context learning step The user's behavior pattern And providing personal adaptive context acquisition method characterized by obtaining a personal adaptive context.
According to this method, by using a user-specific function, a user-specific feature can be learned, and based on the learning result, a personal adaptive context that can grasp the user's behavior in detail and accurately Can be earned.

  According to the present invention, since the mobile terminal learns the correspondence between the feature parameter obtained by developing the sensor data with the user-specific function and the behavior data of the user, the mobile terminal has a unique feature specific to the user. Can learn. For this reason, based on this learning result, it is possible to obtain a personal adaptive context capable of grasping individual behavior in detail and accurately.

Next, embodiments of the present invention will be described with reference to the drawings. In each drawing referred to in the following description, the same reference numerals are given to the same parts as in the other drawings.
FIG. 1 is a block diagram showing a hardware and software configuration of portable terminal 1 according to the embodiment of the present invention.
As shown in the figure, the mobile terminal 10 includes a central processing unit (CPU) 11 that executes a program and controls the entire mobile terminal 10, and a static random access memory (SRAM), a dynamic random access memory (DRAM), and the like. Volatile memory 12, a non-volatile memory 13 composed of a flash memory, etc., and a communication interface 14 for controlling communication via a mobile communication network, and are configured as hardware of a general mobile phone Hardware configuration.

Further, the mobile terminal 10 is equipped with a sensor 15 and a user action history acquisition device 16.
The sensor 15 acquires information about the user such as the user's biological information and environmental information around the user. The sensor 15 includes a plurality of sensors. The plurality of sensors can be classified into a biological information measurement sensor 151 and an environmental information measurement sensor 152. The biological information measurement sensor 151 is a sensor that measures a user's pulse, a sensor that measures a user's body temperature, a sensor that measures sweating conditions (skin resistance), a captured image, a GPS (Global Positioning System), and the like. It includes a sensor that measures the position and orientation of the camera, a sensor that acquires voice information of the user, and the like.

The environmental information measurement sensor 152 is a sensor group that measures environmental information such as temperature and brightness of people and things around the user. Specifically, the environmental information measurement sensor 152 includes a human sensor, an illuminance sensor, an air temperature sensor, a humidity sensor, a noise measurement sensor, and the like.
Further, the plurality of sensors constituting the sensor 15 can be classified into “analog-type” sensors and “on-off-type” sensors from different viewpoints. An “analog system” sensor is a sensor that acquires information as an analog signal, such as a sensor that acquires image information, a sensor that acquires audio information, and a pulse sensor. An “on-off system” sensor is a sensor that reacts only when there is a change in the presence or light quantity of a person, such as a human sensor or an illuminance sensor.

  The user behavior history acquisition device 16 is a device that acquires the behavior history of the user who owns the mobile terminal 10, and is configured of, for example, an IC chip. The user action history acquisition device 16 stores an action history acquisition program 162 for acquiring user action data and a user action history DB (database) 161 for storing the acquired action data. . The user action history acquisition device 16 acquires a service use history in the mobile terminal 10 as user action data, and accumulates the acquired user action data in the user action history DB 161. The service use history includes a history of accessing the Internet from the mobile terminal 10 and using a website (access date / time, URL (Uniform Resource Locator)), history of sending / receiving e-mail (transmission / reception date / time, destination email address, subject) , A purchase history (purchase date and time, purchase store, purchased goods, amount) of electronic money and a call history (call date and time, destination telephone number) are included.

  Various software is stored in the nonvolatile memory 13 of the portable terminal 10. The various software includes a personal adaptive context acquisition program characteristic of the present invention. This personal adaptive context acquisition program is composed of a plurality of program modules. The plurality of program modules include a sensor information division function module 131 for determining a time window of time series data obtained from each sensor based on event (change) information of sensor data obtained from a certain sensor, A user specific function calculation function module 132 for finding a basis function (hereinafter referred to as “user specific function”) suitable for analyzing a user specific pattern from sensor data obtained from a sensor, and divided for each time window Sensor data expansion function module 133 for expressing sensor data using a user specific function as a basis function, and feature parameters (expansion coefficients) obtained by expressing the sensor data using a user specific function as a basis function are input data. User service usage history acquired by the user action history acquisition device 16 As output data to represent the behavior of the user, the individual adaptive context learning module 134 for learning the relationship between the two, are included.

  Further, the software stored in the nonvolatile memory 13 of the mobile terminal 10 includes a database for storing various data. The database is data in which feature parameters are summarized using a method such as a user specific function DB 135 for storing the calculated user specific function and a SOM (Self-Organizing Maps). There is a sensor information summary data DB 136 for storing sensor information feature data, and a learned neural network DB 137 for storing a learning result of a correspondence relationship between sensor information feature data and user behavior data.

Data is temporarily stored in the volatile memory 12. For example, sensor data acquired by the sensor 15 is accumulated in the volatile memory 12.
The CPU 11 of the mobile terminal 10 executes the program modules 131, 132, 133, and 134 that constitute the personal adaptive context acquisition program described above, whereby the functional configuration shown in FIG.

  The sensor data dividing unit 101 divides sensor data acquired by each sensor constituting the sensor 15 for each event. Here, as a method of dividing the sensor data for each event, for example, a method of dividing the sensor data acquired by each sensor every time a change exceeding a threshold occurs in the sensor data acquired from a specific sensor. Conceivable. As another method, every time sensor data acquired by an “on-off system” sensor changes from 1 (on) to 0 (off), or changes from 0 (off) to 1 (on). A method of dividing the sensor data acquired by each sensor every time it is performed can be considered.

The user specific function calculating unit 102 calculates a user specific function for approximating and expressing the sensor data divided by the sensor data dividing unit 101. The user specific function calculation unit 102 stores the calculated user specific function in the user specific function DB 135. As a method for calculating the user eigenfunction, a fractal image compression algorithm method for calculating a self-similarity function may be used, and several localized wave functions such as wavelet transforms are prepared, and those stations are prepared. An expansion method using a standing wave function may be used.
The sensor data expansion unit 103 expands the sensor data divided by the sensor data division unit 101 using the user specific function calculated by the user specific function calculation unit 102.

The personal adaptive context learning unit 104 includes the feature parameters obtained by the sensor data development unit 103 developing the sensor data, and the user behavior acquired by the user behavior history acquisition device 16 when the sensor data is acquired. Learn correspondence with data. Then, the personal adaptive context learning unit 104 stores the learning result in the learned neural network DB 137. As a specific learning method, for example, the personal adaptive context learning unit 104 uses a feature parameter obtained by development as input data, and generates an SOM using an unsupervised competitive model. Then, the personal adaptive context learning unit 104 learns the correspondence between the output data of the generated self-organizing map and the user's action data using a neural network model. The learning method is not limited to this, and other known learning models can be used. For example, a Markov model may be used.
The personal adaptive context acquisition unit 105 acquires a user behavior pattern corresponding to the sensor data sequentially acquired by the sensor 15 as a personal adaptive context based on the learning result by the personal adaptive context learning unit 104.

Next, an operation example will be described with reference to FIGS. FIG. 3 is a flowchart showing a procedure of the personal adaptive context acquisition process performed by the mobile terminal 10. FIG. 4 is a diagram illustrating an example of service provision using the personal adaptive context acquired by the mobile terminal 10.
In the following description, the following symbols e _all , e _analog , e _on-off , C _all , C _analog , C _on-off , and mα represent vectors. Among the sensors constituting the sensor 15, there are N _analog (N _analog is a natural number) total number of “analog type” sensors, and N _on-off (N _on-off ) total number of “on-off type” sensors. Is a natural number).

  The user behavior history acquisition device 16 captures service usage history as user behavior data each time a user sends and receives emails using the mobile terminal 10, accesses the Internet, or purchases products with electronic money. Accumulate in the action history DB 161 (step S101). Further, the “analog system” sensor that constitutes the sensor 15 sequentially captures time series data and stores it in the volatile memory 12 (step S102). Further, the “on-off system” sensor sequentially captures on-off data (data representing 0 (on) or 1 (off)) and stores it in the volatile memory 12 (step S103).

The mobile terminal 10 extracts the i-th (i = 1, 2,..., N _analog ) stored in the volatile memory 12 by the “analog-type” sensor in order to extract user-specific features from each sensor data. Based on the sensor data Sa ⁽ⁱ⁾ (t), a user specific function suitable for extracting a user specific signal pattern is calculated.
Specifically, first, the sensor data dividing unit 101 of the mobile terminal 10 performs j-th (j = 1, 2,..., N _on-off ) sensor data accumulated by the “on-off system” sensor. Based on the time when Sd ^(j) (t) changes from 1 (on) to 0 (off) or from 0 (off) to 1 (on), obtained from an “analog system” sensor The sensor data Sa ⁽ⁱ⁾ (t) is divided (step S104).

Then, the user-specific function calculation unit 102 calculates a basis function that can be approximated for each of the divided sensor data Sa ⁽ⁱ⁾ (t) (t _s ≦ t ≦ t _e ), and calculates the basis function. It is assumed that the eigenfunction U ⁽ⁱ⁾ _k (t) (k = 1, 2,..., N ⁽ⁱ⁾ _analog ) (step S105). As a method for calculating the basis function, here, a known fractal image compression algorithm is used, and it is expressed by a plurality of attractor parameters as corresponding to a user-specific function and a coefficient obtained by the decomposition. For example, the divided sections t _s ≦ t ≦ t _e at Sa by converting the sensor data Sa ^{(i) (t)} in a two-dimensional image data format _{^{(i) (t x, t}} y) is obtained, this fractal An image compression algorithm can be applied. In this case, the user specific function calculation unit 102 converts the attractor parameter (compression code string) C ⁽ⁱ⁾ shown in Expression (1 ⁾ into the user specific function U ⁽ⁱ⁾ _k (t) (= U ⁽ⁱ⁾ _k (t _x , T _y )) (k = 1, 2,..., N ⁽ⁱ⁾ _analog ; n ⁽ⁱ⁾ _analog is a natural number) and is registered in the user specific function DB 135.

The user specific function calculation unit 102 performs the same processing for each sensor, calculates a user specific function U ⁽ⁱ⁾ _k (t _x , t _y ) suitable for extracting a user specific signal, and calculates the result. Accumulate in the user specific function DB 135.
The user eigenfunction calculation unit 102 uses the same technique as the “analog system” sensor for the sensor data Sd ^(j) (t) acquired from the “on-off system” sensor, and uses the user eigenfunction U ^{(j )} _k (t) (= U ^(j) _k (t _x , t _y )) (k = 1, 2,..., n ^(j) _on-off ; n ^(j) _on-off is a natural number) calculate.

Next, sensor information feature data mα is generated using the user specific function U ⁽ⁱ⁾ _k (t _x , t _y ). Specifically, first, the sensor data development unit 103 uses the user-specific function U ⁽ⁱ⁾ _k (t _x , t _y ) for each sensor to divide the sensor data Sa ⁽ⁱ⁾ (t) (t _s ≦ t ≦ t _e) (already converted into two-dimensional image data format) approximates, expressed by a vector e _analog representing a difference between an actual value to the equation (2).

For each “on-off system” sensor, the sensor data development unit 103 also sets the current on-off state to the vector e _on-off = (0, 1, 0,..., 1, 0) (“0 “,“ 1 ”represents an example of an on-off state of each sensor). Then, the sensor data development unit 103 generates e _all shown in Expression (4) as a vector representing the difference from the actual value (step S106).

Note that e _all can be expressed only by e _{analog in} the sense of a vector representing a difference from an actual value.
The personal adaptive context learning unit 104 uses the feature parameters (e _all shown in equation (4) and C _all shown in equation (5)) obtained by developing sensor data in this way as competitive learning. Is performed (step S107).

Here, the SOM is obtained by the known Kohonen competitive learning model. The personal adaptive context learning unit 104 stores the output node mα constituting this SOM in the sensor information summary data DB 136 as sensor information feature data.
Further, the personal adaptive context learning unit 104 receives the sensor information feature data mα obtained as described above, and outputs the user behavior data obtained from the user behavior history DB 161 of the user as a known BP (Back-propagation) learning. The correspondence between the two is learned by a neural network based on the rule (step S108). Here, the output format of the user behavior data is the user for each service usage content such as “Access to XXX site”, “Send email to XXX”, and “Purchase goods with △△△”. An action node P _i is prepared, and 1 is assigned if the user is actually taking action, and 0 is assigned if the action is not actually taken. The personal adaptive context learning unit 104 stores the learning result (learned neural network) at this time in the learned neural network DB 137.

A series of processes up to the above are always performed, and each DB is updated.
In parallel with this series of processing, the personal adaptive context acquisition unit 105 acquires a personal adaptive context based on the learning result of the above processing. Specifically, the sensor data Sa ⁽ⁱ⁾ (t) and Sd ^(j) (t) captured by the “analog system” and “on-off system” sensors are the user-specific function U ⁽ⁱ⁾ described above. Based on (t), it is expanded into feature parameters e _all and C _all and further converted into sensor information feature data mα. The personal adaptive context acquisition unit 105 acquires a personal adaptive context corresponding to the sensor information feature data mα based on the learned neural network DB 137 (step S109). Specifically, when the sensor information feature data mα is applied as input data to the learning result stored in the learned neural network DB 137, the user service usage pattern becomes the user behavior pattern (P ₁ , P ₂ ,. P _i ). The action pattern here is expressed by attribute information representing each action of the user obtained by the user action history acquisition device 16 and the probability of the occurrence. For example, if the user action of event 1 is a search for information on a movie site, the action pattern is attribute information “URL: http: //www.EIGA.co.jp/xxx / ...” is represented by a probability P ₁ that occurs, if the user behavior of the event 2 has sent a greeting mail to Taro, behavior pattern is "mail: to: Taro@zzz.co.jp, Subject: " Hello! "" attribute information and it is represented by a probability P ₂ that occurs that, further, if the user behavior of the events i bought a jacket of 1,000 yen at the store of XYZ using electronic money, behavior pattern is "e -cash: Store: XYZ, Product: Jacket, Price: 1,000 Yen "and the probability P _i of the occurrence. This expresses the probability that each behavior pattern can occur, that is, the probability of the behavior that the user may take at this point in time.

  The existing service providing system 20 having a CRM (Customer Relationship Management) database 30 shown in FIG. 4 uses a CRM database to provide services suitable for user behavior predicted from the personal adaptive context acquired by the mobile terminal 10. Search from 30. Thereby, the service providing system 20 can find a service suitable for the user (step S110). The service providing system 20 provides the discovered service to the mobile terminal 10 (step S111). For example, if the personal adaptive context indicates that the user is currently accessing a movie-related site, the service providing system 20 transmits information related to ticket reservation to the mobile terminal 10. In addition, if the personal adaptive context indicates that the user is currently accessing a trendy fashion site and the user has a high probability of going out to purchase clothes after accessing the site, the service providing system 20 transmits clothes bargain information to the portable terminal 10. In this way, the mobile terminal 10 can acquire a service corresponding to the personal adaptive context from the service providing system 20.

  It is also possible to find a service corresponding to a personal adaptive context by using an existing ontology method. In this way, the mobile terminal 10 can learn and acquire a context adapted to the individual user. The service providing system 20 can provide the user with a service suitable for the user based on the context acquired by the mobile terminal 10.

  As described above, the mobile terminal 10 calculates a user-specific function for each user, for each sensor, and for each event, and obtains a characteristic parameter obtained by developing sensor data using the user-specific function and user action history acquisition. Since the correspondence relationship with the behavior data of the user obtained by the device 16 is learned, it is possible to learn the features unique to the user and to acquire a context adapted to the individual based on the learning result. .

In addition, since the sensor data is automatically divided for each event and the user behavior history is acquired by the user behavior history acquisition device 16, knowledge processing and subjective context definition are not required. Only by carrying and using the mobile terminal 10 as usual, it is possible to learn and acquire a personal adaptive context.
Further, by using a user specific function, sensor data can be converted into a typical user characteristic parameter C _all (a characteristic parameter based on the user specific function) and an exceptional characteristic parameter e _all (the difference between the sensor data and the user specific function). Therefore, a large number of user-specific features can be extracted from the sensor data, and the feature parameters and user behavior data can be associated with each other accurately.

Further, based on the degree of difference between the sensor data and the user specific function, it is possible to interpret whether or not the behavior pattern is normal. For example, if sensor data acquired every day in a specific time zone can always be expressed with the same user-specific function, the user always takes the same action in a specific time zone and there is a habit of behavior I understand. On the other hand, when sensor data acquired at the same specific time zone is expressed by a usual user-specific function on one day, if e _all becomes large, it can be seen that the user has taken a different behavior.

  The acquired personal adaptive context is information that can be an element of the existing CRM procedure (information expressing the availability of services (that is, behavior patterns) and the relationship of each service with probability and correlation). Therefore, there is an affinity with the service providing system 20 based on the existing CRM. Therefore, the service providing system 20 can easily find and provide a service suitable for the user using the personal adaptive context and the CRM database 30.

In the above-described embodiment, the sensor 15 has been described as being mounted on the mobile terminal 10, but the present invention is not limited to this. For example, the band on which the sensor 15 is mounted is placed on a human arm or head. It may be wound.
Further, in the above-described embodiment, it has been described that the user action history acquisition device 16 configured with an IC chip is mounted on the mobile terminal 10, but a program for acquiring a user action history is stored in the nonvolatile memory 13. The function of the user action history acquisition device 16 may be realized by storing the function.
The mobile terminal 10 is not limited to a mobile phone, and may be any device that can be carried by a user and can use various services. For example, a PDA (Personal Digital Assistance) or PHS (Personal Handyphone System) may be used.

  It can be used to recognize an individual situation and can be used to provide a service according to the recognized individual situation. For example, it can be used for personal health care, service navigation based on CRM, and the like.

It is a block diagram which shows the hardware and software structure of the portable terminal which concerns on embodiment of this invention. It is a block diagram which shows the function structure of the portable terminal which concerns on the same embodiment. It is a flowchart which shows the procedure of the personal adaptive type context acquisition process which the portable terminal which concerns on the embodiment performs. It is a figure which shows an example of the service provision using the personal adaptation type | mold context acquired by the portable terminal which concerns on the embodiment.

Explanation of symbols

10 Mobile terminal 11 CPU
DESCRIPTION OF SYMBOLS 12 Volatile memory 13 Non-volatile memory 131 Sensor information division | segmentation function module 132 User specific function calculation function module 133 Sensor data expansion | deployment function module 134 Personal adaptive type context learning function module 135 User specific function DB
136 Sensor information summary data DB
137 Trained neural network DB
14 Communication Interface 15 Sensor 151 Biological Information Measurement Sensor 152 Environmental Information Measurement Sensor 16 User Action History Acquisition Device 161 User Action History DB
162 Action history acquisition program 101 Sensor data division unit 102 User specific function calculation unit 103 Sensor data development unit 104 Personal adaptive context learning unit 105 Personal adaptive context acquisition unit 20 Service providing system 30 CRM database

Claims (5)

Sensor data dividing means for dividing sensor data indicating information about the user acquired by the sensor;
User specific function calculating means for calculating a user specific function for expressing the sensor data divided by the sensor data dividing means;
Sensor data expansion means for expanding the sensor data divided by the sensor data division means using the user specific function calculated by the user specific function calculation means;
A portable terminal comprising: a personal adaptive context learning means for learning a correspondence relationship between a feature parameter obtained by the development by the sensor data development means and user behavior data. It further comprises user behavior data storage means for storing service usage history in its own terminal as user behavior data,
The personal adaptive context learning means includes:
The mobile terminal according to claim 1, wherein a correspondence relationship between user behavior data stored in the user behavior data storage unit and the feature parameter is learned. The sensor data expansion means is
The mobile terminal according to claim 1, wherein the mobile terminal is developed into a characteristic parameter based on the user specific function and a characteristic parameter based on a difference between the sensor data and the user specific function. Based on a learning result by the personal adaptive context learning means, further comprising a personal adaptive context acquisition means for acquiring a user behavior pattern corresponding to the sensor data acquired by the sensor as a personal adaptive context;
The mobile terminal according to any one of claims 1 to 3, wherein the mobile terminal receives a service corresponding to the personal adaptive context acquired by the personal adaptive context acquisition means. A sensor data dividing step for dividing sensor data indicating information about the user acquired by the sensor;
A user specific function calculating step for calculating a user specific function for expressing the sensor data divided in the sensor data dividing step;
Sensor data expansion step for expanding the sensor data using the user specific function calculated in the user specific function calculation step;
A personal adaptive context learning step for learning a correspondence relationship between the feature parameter obtained by the expansion in the sensor data expansion step and the behavior data of the user,
A personal adaptive context acquisition method, wherein a user behavior pattern corresponding to sensor data acquired by the sensor is acquired as a personal adaptive context based on a learning result in the personal adaptive context learning step.

Images