JP3822522B2 - Face-to-face support system using location information - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a system that supports communication between users of a portable information terminal.
[0002]
[Prior art]
In recent years, computer systems using portable information terminals such as mobile phones, PHS, and PDAs have become widespread.
As shown in FIG. 1, for example, such a computer system includes a portable information terminal 112 at hand of each user who is a user of the system and a server system 120 that performs overall data acquisition and processing. . The server system 120 includes a management server 122 that stores a system control program and a database 128, a WWW server 124 that stores a Web page, a mail server 126 that manages e-mail, and the like. The user can send and receive e-mails and refer to Web pages by communicating with the server system 120 using the portable information terminal 112 or a personal computer.
In addition, some portable information terminals 112 have a GPS function. Using this function, the current position information of the user is grasped, the surrounding information is notified to the user, and the user is guided to the destination. The system is also becoming popular.
In addition, a system in which a user registers his / her schedule in the database 128 and the management server 122 manages the schedule has become common. In the case of such a system, each user accesses the WWW server 124, displays a schedule creation page as shown in FIG. 2 on the screen of his / her portable information terminal 112, and inputs the schedule. For example, the date of the schedule is entered in the date entry field 221, the schedule content such as “meeting” is entered in the errand entry field 222, and the location of the schedule such as “first meeting room” is entered in the place entry field 223. In the column 224, the start time and end time of the schedule are input, and in the importance input column 225, the importance of the schedule is input. When input is made in these input fields and the “Send” button 226 is pressed, the management server 122 receives the input content and registers the schedule in the database 128. Later, the user who has input the schedule can refer to the input schedule using the portable information terminal.
Furthermore, if the conventional techniques as described above are used, information such as each user's schedule and current position can be shared between users, and as a result, activation of communication between users can be expected.
[0003]
[Problems to be solved by the invention]
However, in the conventional computer system, the location information that can be shared by the user is limited to the current location information. For this reason, even if a certain user wishes to meet another user in the future, the user's position at the date and time when the user wishes to meet cannot be known. Also, it is impossible to know when there is a high possibility of meeting. As a result, it is considered that the opportunity for communication between users is lost.
Therefore, the face-to-face support system according to the present invention predicts the future position of a user who is desired to meet when a certain user wants to meet another user in the future, and allows the user who wants to meet to meet The purpose is to support the determination of whether or not to have the opportunity to meet, and to support the determination of the schedule and place to meet.
[0004]
[Means for Solving the Problems]
In order to achieve the above object, the present invention is a face-to-face support system that has communication means with a portable information terminal capable of transmitting a position and supports communication with a user of the portable information terminal. And receiving location information of each user transmitted from the portable information terminal carried by each user, and storing location information storage means for storing history information of each user's location information; and When the behavior model construction means for constructing the behavior model of each user using the history information of the accumulated position information and the request for meeting with the specific user and the desired meeting date and time are input, the construction is performed. And a position prediction unit that predicts the position of the specific user at the desired meeting date and time and outputs the result using the behavior model of each user. A.
The behavior model construction unit may divide the position information for each day of the week and each time zone and construct a behavior model corresponding to each day of the week and each time zone.
Furthermore, it has a schedule storage means for storing the schedule information of the user of the portable information terminal, and the position prediction means further performs prediction using the schedule registered in the database by the specific user. May be.
The position prediction means may further perform prediction using position information recently acquired by the specific user.
A program that causes a computer system to construct the above-described meeting support system and a recording medium that records the program are also the present invention.
[0005]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the face-to-face support system will be described in detail with reference to the drawings.
<Configuration of this embodiment>
The configuration of the face-to-face support system of the present embodiment is the same as the system configuration diagram of FIG. Note that the face-to-face support system of the present embodiment can manage users in units of groups and support communication between users in the group, for example. A group is a group of users such as university circles and seminars, for example, and a single user can belong to a plurality of groups.
[0006]
<Operation of this embodiment>
Next, operation | movement of the meeting assistance system of this embodiment is demonstrated.
In the face-to-face support system of this embodiment, the position information of each user is acquired, and an action model is constructed using the history of the acquired position information. And when there is a request for face-to-face support from the user, the position is predicted using the constructed behavior model, schedule, and the latest position information, and the predicted position is presented to the user, so that communication between users can be performed. I support.
Among the above, the manner in which the face-to-face support system of the present embodiment manages the schedule has already been described as the prior art, so here, acquisition of position information, construction of an action model, and position prediction will be described in order.
[0007]
(Acquisition of location information)
First, a state in which the face-to-face support system of the present embodiment acquires user position information will be described.
Acquisition of the user's position information is performed using the portable information terminal 112 carried by the user. Specifically, for example, a user's position is acquired by using a GPS function mounted on a mobile phone, a position acquisition function by a PHS base station, and the management server 122 receives the acquired position. Do. Further, the acquisition of the position information is periodically performed, for example, every 15 minutes. Furthermore, you may provide the function acquired as needed as needed. The user's position is registered and stored in the database 128 as information such as a place name and latitude / longitude, for example. The location information accumulated here is used in the behavior model construction described later.
[0008]
(Building a behavior model)
Next, the manner in which the face-to-face support system of the present embodiment constructs a user behavior model will be described using the flowchart of FIG.
First, a user's action pattern is grasped using position information accumulated in database 128 (S311). Since the location information is periodically acquired, for example, every 15 minutes, if the same location information continues from the history of the location information of the user accumulated in the database 128, the “stop” state, location information When is switched, it is determined that the state is “move”. Then, when the user's state is switched from “Stop” to “Move”, and from “Move” to “Stop” again, the first position information to the last position information in that period are grasped as one action pattern. To do.
[0009]
Next, the grasped action pattern is narrowed down (S312). This is because if all of the grasped behavior patterns are used at the time of constructing a behavior model to be described later, there may be a problem such as a long processing time due to the large number of behavior patterns. Specifically, for example, action patterns are narrowed down by the following methods (A) to (D). The action pattern narrowing down method is not limited to (A) to (D), and other methods may be used.
(A) The grasped behavior pattern is selected in descending order of the number of times the user has taken the behavior pattern until a threshold value (for example, 10) is reached.
(B) The number of grasped behavior patterns is added in the descending order of the number of times the user has taken the behavior pattern, and is selected until the threshold value (for example, 70%) is reached among the total number of behavior patterns.
(C) The grasped behavior pattern is selected in order of the longest time when the user took the behavior pattern until a threshold value (for example, 10) is reached.
(D) The time of the grasped behavior pattern is added in the order of the longest time when the user took the behavior pattern, and is selected until a threshold value (for example, 70%) is reached in the entire time.
[0010]
Next, a behavior model of the user is constructed using the narrowed behavior pattern (S313). For example, the behavior model is constructed by the following methods (1) and (2). In addition, the construction method of an action model is not limited to (1) and (2), You may use another method.
(1) Construction of a probability model using a genetic algorithm In this method, a behavior model of a user is constructed as a probability model by obtaining the probability of transition from each behavior pattern to another behavior pattern using a genetic algorithm. To do.
First, a population of individuals expressing transition probabilities as a matrix is generated. For example, a certain number (for example, 1000) of individual populations are generated using random numbers.
Next, the individual is evaluated, and an evaluation value indicating how much each individual can model the behavior of the user is calculated. When evaluating an individual, the behavior pattern at a certain time of a user is input, the probability of which behavior pattern is taken at the next time is calculated using the transition probability matrix expressed by the individual, and this is actually used. Each individual is evaluated by collating with the data. As actual data used here, for example, a certain number of behavior patterns at various times are selected as test data using random numbers from a history of user position information.
A process called crossover or mutation in a genetic algorithm by selecting two or one individual from a population using the probability according to the evaluation value of each individual, the order of evaluation values, the size of evaluation values, etc. To generate new individuals. This process is repeated until the number of the population is the same, and the next generation individual population is generated.
For each individual of the newly generated population, the individual is evaluated again. This operation is repeated, and when the number of generations reaches the threshold value or when the evaluation value of the individual reaches the threshold value, the learning cycle by the genetic algorithm is terminated.
[0011]
(2) Construction of a descriptive model using genetic programming In the method (1) described above, the behavior model is constructed on the assumption that the user's behavior is related only to the previous behavior. In the method (2), the behavior model is constructed on the assumption that the behavior of the user is related to the behavior that goes back from the previous behavior.
First, like the genetic algorithm of (1), a certain number (for example, 1000) of individual populations are generated using random numbers. Here, the term “individual” refers to a program that expresses a vector, for example, a program that uses as an argument a numerical expression of position information as latitude and longitude.
Next, the individual is evaluated, and an evaluation value indicating how much each individual can model the behavior of the user is calculated. When an individual is evaluated, an action pattern of a user at a certain time and n patterns going back from that time are input, and what action pattern is taken at the next time using a program represented by the individual Each individual is evaluated by calculating a vector and comparing it with actual data.
A process called crossover or mutation in genetic programming by selecting two or one individual from the population using the probability according to the evaluation value of each individual, the order of the evaluation values, the magnitude of the evaluation values, etc. To generate new individuals. This process is repeated until the number of the population is the same, and the next generation individual population is generated.
For each individual of the newly generated population, the individual is evaluated again. This operation is repeated, and when the number of generations reaches the threshold value or when the evaluation value of the individual reaches the threshold value, the learning cycle by genetic programming is completed.
[0012]
In the face-to-face support system of the present embodiment, an action model is constructed by the methods (1) and (2) described above, and the constructed action model is registered in the database 128. Since the user's position information is always accumulated, the process shown in the flowchart of FIG. 3A is periodically performed (for example, once a day) to update the behavior model.
In addition, since it is considered that the user's behavior pattern varies depending on the day of the week and the time zone, when constructing the behavior model, the behavior pattern is divided into the day of the week and the time zone, and the behavior model corresponding to each time zone of each day of the week Also good to build. Thereby, when predicting the position of the user, it is possible to select an action model on the same day of the week and the same time zone as the date and time for which the position is to be predicted, and more accurate prediction is possible.
[0013]
(Prediction of position)
Next, the manner in which the face-to-face support system of the present embodiment predicts the position of the user will be described using the flowchart of FIG.
Here, as an example, a case where the user A desires to have an opportunity to meet the user B will be described. It is assumed that the user B always carries a portable information terminal that can transmit the above-described position information, and the behavior model of the user B is constructed in the database 128.
[0014]
User A uses the same Web page as the above-described schedule creation page in FIG. This input from the user A can be performed from a terminal such as a personal computer in addition to the portable information terminal 112. In the face-to-face support system according to the present embodiment, the day to be met or the day after the next day is input as the face-to-face desired date. Any time zone can be entered in. The management server 122 receives the input desired meeting date and time (S321).
Next, the meeting support system searches the database 128 for the schedule of the user B at the desired meeting date and time (S322).
When there is a schedule at the desired meeting date and time, the location of the schedule is transmitted as it is to the user A as the predicted position of the user B (S327).
If there is no schedule at the desired meeting date and time, the user B's schedule up to the desired meeting date and time is searched from the database 128 (S323). If the desired face-to-face date is today, the schedule from the current time to the desired face-to-face time is searched, and if the desired face-to-face date is after the next day, the schedule up to the desired face-to-face time zone is searched. In addition, since there may be a case where there are a plurality of schedules corresponding to the above, a schedule having a time closest to the desired meeting time or the desired meeting time zone is searched.
When there is a schedule up to the desired meeting date and time, based on the end time and location of the schedule, the database 128 is searched to select the user B's behavior model, and the position of the user B at the desired meeting date and time is predicted (S326). ). A specific prediction method in this case will be described later.
[0015]
If there is no schedule until the meeting date and time, it is determined whether the meeting date is today (S324).
If the desired meeting date is today, the latest location information of the user B is referred from the database 128 (S325), the database 128 is searched based on the location information, and the behavior model of the user B is selected. The position of the user B at the date and time is predicted (S326). A specific prediction method in this case will be described later.
If the desired meeting date is on or after the next day, the database 128 is searched to select the behavior model of the user B on the same day of the week and the same time zone as the desired meeting date and time, and the position of the user B at the desired meeting date and time is predicted (S326). ). In this case, it is only necessary to select an action model corresponding to the day of the week and the time zone of the desired meeting date and time, and it is not necessary to use a prediction method described later.
Finally, the predicted position of the user B is transmitted to the portable information terminal 112 of the user A or a personal computer (S327). User A can refer to the predicted position of user B on the same Web page as the schedule creation page of FIG.
[0016]
In the flowchart of FIG. 3B, the prediction of the position of the user B in S326 is specifically performed by the following method.
When the behavior model of user B is constructed as the probability model of (1) described above, the end time of the schedule closest to the desired meeting date and time (the desired meeting date is today and there is no schedule until the desired meeting date and time) In this case, the number n of the transition probability matrix is calculated from the difference between the current time and the desired date and time, and the location of the schedule closest to the desired date and time (the desired date is today and the desired date and time) If there is no schedule, the position of user B after n steps from the current position) is predicted. In this case, a plurality of pieces of position information of the user B at the desired meeting date and time are predicted, and the probability that the user B is in each predicted position is presented to the user A together with each predicted position.
When the user B's behavior model is constructed as the descriptive model described in (2) above, the next position information is predicted from the current position information of the user B and i pieces of position information retroactive to the present. To do. The next position information is further predicted from the predicted position information, the current position information, and the i-1 pieces of position information going back from the present. By repeating this operation, the position information of the user B after n steps is predicted from the difference between the current time and the desired meeting time. In this case, one piece of position information of the user B at the desired meeting date and time is predicted, and the probability that the user B is at the position is presented to the user A as a probability together with the predicted position.
[0017]
【The invention's effect】
The face-to-face support system of the present invention predicts the position at the face-to-face desired date and time of a user who is desired to face when a certain user wants to face another user in the future, and wants to meet By presenting it to the user on the side, it is possible to support the determination of whether or not to have the opportunity to meet, or to support the determination of the schedule and place to meet, and to activate communication between users it can.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating a configuration of a meeting support system according to an embodiment.
FIG. 2 is a diagram illustrating an example of a schedule creation page of a face-to-face support system.
FIG. 3A is a flowchart for explaining how the face-to-face support system constructs a user behavior model.
(B) It is the flowchart explaining a mode that a facing support system estimates a user's positional information.

Claims (6)

A face-to-face support system having communication means with a portable information terminal capable of transmitting a position and supporting communication with a user of the portable information terminal;
Position information storage means for receiving position information of each user transmitted by the portable information terminal carried by each user and storing history information of the position information of each user;
Action model construction means for constructing an action model of each user using history information of the position information accumulated in the position information accumulation means;
When the user wishes to meet a specific user and inputs the desired date and time of the meeting, the position of the specific user at the desired date and time of the meeting is predicted using the constructed behavior model of each user. And a face prediction system for outputting the result. The face-to-face support system according to claim 1.
The face-to-face support system, wherein the behavior model construction means divides the position information for each day of the week and each time zone and constructs a behavior model corresponding to each day of the week and each time zone. In the face-to-face support system according to claim 1 or 2,
Furthermore, it has schedule accumulation means for accumulating schedule information of users of portable information terminals,
The face-to-face support system, wherein the position prediction means further performs prediction using a schedule registered by the specific user in the database. In the meeting assistance system in any one of Claims 1-3,
The face-to-face support system, wherein the position prediction means further performs prediction using position information recently acquired by the specific user. The recording medium which recorded the program which makes a computer system build the meeting assistance system in any one of Claims 1-4. The program which makes a computer system construct the meeting assistance system in any one of Claims 1-4.

Images