KR101328328B1 - Apparatus and method for estimating location using log data - Google Patents

Abstract

The present invention relates to an apparatus and method for estimating a place using log data. The apparatus for estimating a place according to an embodiment of the present invention includes a log data collection unit for collecting log data from a user terminal and a user by learning the log data. And a place estimating model generating unit for generating a place estimating model for the data; and a place estimating unit for analyzing log data collected from the user terminal and estimating a current place of the user using the place estimating model.
Accordingly, the user's behavior pattern can be analyzed more accurately, and the current location can be accurately estimated.

Description

Apparatus and method for location estimation using log data {APPARATUS AND METHOD FOR ESTIMATING LOCATION USING LOG DATA}

The present invention relates to an apparatus and method for estimating a place using log data, and more particularly, a technique for estimating a place where a current user is located by analyzing a usage history of a user terminal.

Due to the rapid proliferation of smart phones, the situational awareness service using various sensors embedded in smart phones is being advanced. As a result, location information services using GPS sensors such as LBS (Location Based Service) can be easily and freely used at low cost. However, the location recognition method using only the GPS sensor has an accuracy within the error range of 5 to 20 m outdoors, but there is a problem in that the location information cannot be provided due to the attenuation of the GPS signal strength in the indoor, urban canyon, or tunnel. .

In order to solve this problem, a method for compensating location information through a complex positioning method using a base station information of a mobile phone, Wi-Fi AP (Acess Point) information, etc. has been researched and developed. In addition, the government also announced legislation to improve the domestic LBS market and secure competitiveness. In order to improve the accuracy of location information, we are pursuing mandatory installation of GPS functions in mobile phone terminals, and expanding infrastructure for eliminating shaded areas such as underground and indoor spaces.Telecommunications companies, manufacturers, portal operators, etc. A bill is being made to jointly build and operate a database of information on the location of Internet Wi-Fi APs. However, the initial cost of constructing the infrastructure is enormous, and the problem of returning relatively inaccurate place information can cause confusion in inferring the correct place.

Meanwhile, with the advent of Web 2.0, which emphasizes user participation and sharing, social network services (SNS), which focuses on human relationships, are growing rapidly. While SNS was initially provided mainly on the web, the size of SNS combined with mobile is increasing due to changes in the mobile environment such as the spread of 3G networks and the popularization of smartphones. In particular, the spread of smart phones that can recognize a user's current location using GPS, Bluetooth and Wi-Fi has emerged a location-based SNS utilizing the user's location information in the existing SNS.

As a study related to location-based SNS using a smartphone, a study has been conducted on a service for sharing location information between users by using relationship information with other users set by the user. However, in such a service, a user must directly set all social relationship information with another user, and there is a problem in that a service cannot be provided for another user who has not established a relationship.

Accordingly, the present inventors have studied a technology for estimating a user's place using a user terminal such as a smartphone. The background technology of the present invention is described in Republic of Korea Patent Publication No. 10-2007-0009267 (January 18, 2007).

The technical problem to be achieved by the present invention is to estimate the place where the current user is located by analyzing the usage history of the user terminal, and to estimate the place where the current user is located by grouping the adjacent users.

An apparatus for estimating a place according to an embodiment of the present invention includes a log data collector for collecting log data from a user terminal, a place estimating model generator for learning a log data to generate a place estimating model for a user, and the user. And a place estimator for analyzing the log data collected from the terminal and estimating the current place of the user using the place estimation model.

The log data may include at least one of time, place, and neighbor user information of the user terminal acquired through a position sensor or a proximity sensor mounted on the user terminal.

The apparatus may further include a user group model generator configured to generate a user group model generated by grouping adjacent user information located within a preset range from the user terminal, wherein the place estimator uses the place estimating model. It is possible to estimate a place for the user according to.

In addition, the user terminal may communicate data with another neighboring user terminal in at least one communication method of Bluetooth, Wi-Fi, 3G, 4G.

In addition, the location estimation model may be generated using a naive Bayes classification algorithm, and the user group model may be generated using an EM clustering algorithm.

According to another aspect of the present invention, there is provided a method of estimating a place using log data, the method of estimating a place using log data of a place estimating apparatus, comprising: collecting log data from a user terminal and learning the log data to a user; Generating a location estimation model for the location; and analyzing log data newly collected from the user terminal, and estimating the current location of the user using the location estimation model.

As described above, according to the present invention, the usage history of the user terminal is estimated to estimate the place where the current user is located, and the neighboring user is grouped to estimate the place where the current user is located according to the group, thereby analyzing a more accurate user behavior pattern. And accurately estimate where you are located.

1 is a block diagram of an apparatus for estimating a place using log data according to an embodiment of the present invention;
2 is a flowchart of a place estimating method implemented through the place estimating apparatus according to FIG. 1;
3 is an exemplary diagram for explaining estimating a current place of a user by analyzing log data in the place estimating method according to FIG. 2;
4 illustrates an example of estimating a current place of a user using a user group model generated by grouping adjacent user information in the place estimation method of FIG. 2.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. The terms used are terms selected in consideration of the functions in the embodiments, and the meaning of the terms may vary depending on the user, the intention or the precedent of the operator, and the like. Therefore, the meaning of the terms used in the following embodiments is defined according to the definition when specifically defined in this specification, and unless otherwise defined, it should be interpreted in a sense generally recognized by those skilled in the art.

The apparatus for estimating a place of the present invention may be mounted in the user terminal and may be implemented in the form of a server.

1 is a configuration diagram of a place estimating apparatus using log data according to an embodiment of the present invention, and FIG. 2 is a flowchart of a place estimating method implemented through the place estimating apparatus according to FIG. 1.

1 and 2, a place estimating apparatus 100 according to an embodiment of the present invention may include a log data collector 110, a log DB 115, a place estimation model generator 120, and a place estimator. 130. First, the log data collection unit 110 collects log data from the user terminal and stores the log data in the load DB (S200). The user terminal is a mobile terminal such as a PDA, a PMP, a navigation device, a smartphone, a tablet PC, a laptop, and the user terminal communicates data with another neighboring user terminal through at least one of Bluetooth, Wi-Fi, 3G, and 4G. can do.

The log data is stored in the log DB 115 and includes at least one of time, place, and neighbor user information of the user terminal acquired through a position sensor or a proximity sensor mounted on the user terminal. For example, the log data collecting unit 110 collects log data from the user terminal at predetermined time intervals, and the user terminal stores data through Bluetooth communication with an adjacent user terminal and outputs the data instance command if there is a data instance command. do. The log data may include a record of what time the user terminal communicated with which user terminal, and the type of communication may generate log data even by communication of ID confirmation between the user terminals.

Next, the place estimation model generator 120 learns log data of the user terminal stored in the log DB 115 and generates a place estimation model for the user (S210). The place estimation model analyzes log data collected from a user terminal for a certain period of time, and generates a place estimation model by matching which user terminal the user terminal communicates with each time, and when the adjacent user is determined, matching the place where the user is staying with the user. do.

In addition, the place estimation model generator 120 may generate the place estimation model using a naive bayes classification algorithm. The naive Bayes classification algorithm is for generating a user's location estimation model from log data. The naive Bayes classification algorithm is a classification learning technique based on Bayes Theorem that determines the probability P (C | X) for unknown sample X whose class is unknown and determines the probability P (C | X). This is the way to predict with the largest class. In this case, P (C | X) can be obtained using Equation 1 below.

In Equation 1, C represents a class and X represents an unknown sample. Since P (X) has the same constant value in all classes, the value of P (X | C) * P (C) is needed to compare the probability rankings belonging to the class. The value of P (C) can be obtained through the learning example, and P (X | C) can be calculated as the product of the probability values of each variable that can be obtained from the learning example under the assumption that all variables except class variables are independent of each other. have. P (X | C) can be obtained using Equation 2 below.

Next, the place estimator 130 analyzes the log data newly collected from the user terminal and estimates the user's current place using the place estimation model (S220). The place estimator 130 may estimate a current user's place by using the place estimation model generated by the place estimation model generator 120. It analyzes the log data of the user terminal at the current place and estimates the current place by applying it to the place estimation model. Hereinafter, estimating the current location of the user by analyzing log data will be described in detail with reference to FIG. 3.

FIG. 3 is an exemplary diagram for describing estimating a user's current place by analyzing log data in the place estimation method of FIG. 2.

Referring to FIG. 3, when estimating the current location for user 1, analyzing log data received from the user terminal, it can be seen that user 1 is adjacent to another user whose user ID is 76 at 8:00 on Monday. have. In this case, according to the place estimating model generated by the place estimating model generation unit 120, the place with the user ID 76 at 8:00 on Monday is set as a company, and the place estimating unit 130 is a place where the current user is located. Will be assumed to be the company. In addition, when the user 1 is adjacent to the user whose user ID is 76 at 9:00 on Wednesday, the user's location is estimated as the coffee shop.

Meanwhile, the place estimating apparatus 100 according to another exemplary embodiment of the present invention may further include a user group model generator 140. The user group model generation unit 140 generates a user group model generated by grouping adjacent user information located within a preset range from the user terminal. The user group model refers to a model generated by grouping users and adjacent users. In the case of FIG. 3, the place estimation model is generated by using log data of the entire user, and the estimated place may be different from the current place of the user by applying the same to individual users. Therefore, in order to estimate a more accurate place, the user and the adjacent user are grouped to apply a place estimation algorithm for each group. In this case, the place estimator 130 estimates a place for the user according to the user group model using the place estimation model.

In addition, the group model generator may generate a user group model using an EM clustering algorithm. The EM clustering algorithm is an algorithm for finding the maximum likelihood estimates of parameters in a probability model that depends on invisible latent variables. In the Expectation (E) stage, the expected value of the latent variable is calculated, and Maximization (M) calculates the maximum likelihood estimate of the complements using the given data and the latent variable to which the expected value is assigned. The expected step can be obtained by Equation 3 below, and the maximizing step can be obtained by Equation 4 below.

Given a certain parameter θ ^(t) in Equation 3, we define the expected value Q of the maximum likelihood of the probability distribution for the observable random variable X and the unobservable random variable Z when using the new parameter θ. . In addition, in Equation 4, a new parameter θ ^{(t + 1)} maximizing the expected value Q of the maximum likelihood is repeatedly calculated according to Equation 3, and the position where the values sufficiently converge is calculated.

When the EM clustering algorithm is used, cluster data shown in Table 1 below can be obtained.

group Group 1 Group 2 Group 3 Group 4 Group 5 Number of users One 13 24 51 5

Referring to Table 1, a total of five user groups can be set, and the number of users in each group represents the number of users belonging to the same place through the EM clustering algorithm. For example, in group 1, users with IDs 2, 4, 6, 11, 38, 39, 40, 43, 59, 73, and 82 may have an average frequency of appearing in the vicinity of the target user at 5 ~ 5 per day. Users who are about 10 minutes long are grouped together. When training data is extracted by grouping users having similar behavior patterns, a place estimation model is generated for each group by applying the naive beige classification algorithm described above. The number of place estimation models for each group for the group of Table 1 may be expressed as shown in Table 2 below.

group Group 1 Group 2 Group 3 Group 4 Group 5 Place estimate
Number of models
76
67
789
4
5

4 illustrates an example of estimating a current place of a user using a user group model generated by grouping adjacent user information in the place estimation method of FIG. 2.

Referring to FIG. 4, log data of a corresponding user may be collected from a target user terminal to estimate a current location, and a user group adjacent to the corresponding user may be set using a user group model and a location estimation model. For example, in FIG. 4, if a user is in a location adjacent to user 76 whose ID is 76 at 8:00 on Monday, user and adjacent user 76 are simultaneously included in group 1 and group 2, and the behavioral characteristics of the target user are group 1. If it is determined that the current place of the target user is estimated to be a company. However, if the behavioral characteristics of the target user are determined to be group 2, the current place of the target user is assumed to be a library. Therefore, compared to FIG. 3, the range of the estimated place of the target user may be further divided by grouping the neighboring users according to the behavior characteristics of the neighboring users.

As described above, according to the present invention, the usage history of the user terminal is estimated to estimate the place where the current user is located, and the neighboring user is grouped to estimate the place where the current user is located according to the group, thereby analyzing a more accurate user behavior pattern. And accurately estimate where you are located.

Meanwhile, an embodiment of the present invention may also be implemented in the form of a recording medium including instructions executable by a computer, such as a program module executed by the computer. Computer readable media can be any available media that can be accessed by a computer and includes both volatile and nonvolatile media, removable and non-removable media. In addition, the computer-readable medium may include both computer storage media and communication media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Communication media typically includes any information delivery media, including computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave, or other transport mechanism.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, Therefore, the present invention should be construed as a description of the claims which are intended to cover obvious variations that can be derived from the described embodiments.

100: place estimation device
110: log data collection unit
115: log DB
120: place estimation model generator
130: place estimation unit
140: user group model generation unit

Claims (10)

A log data collecting unit collecting first log data from a user terminal;
A place estimation model generator for learning the first log data and generating a place estimation model for a user using a naive Bayes classification algorithm; And
A place estimating unit configured to analyze second log data collected from the user terminal and estimate a current place of the user using the place estimating model,
The first and second log data is obtained through a position sensor or a proximity sensor mounted on the user terminal,
The first log data includes at least one of time, place, and neighbor user information of the user terminal, and the second log data is a place using log data including at least one of time, neighbor user information of the user terminal. Estimation device. delete The method of claim 1,
And a user group model generator configured to group adjacent user information located within a preset range from the user terminal and generate a user group model using the naive Bayes classification algorithm and an EM clustering algorithm.
And the place estimating unit uses log data to estimate a place for the user according to the user group model using the place estimating model. The method of claim 1,
The user terminal is a place estimation device using the log data for communicating data with another adjacent user terminal in at least one communication method of Bluetooth, Wi-Fi, 3G, 4G. delete In the place estimation method using the log data of the place estimation device,
Collecting first log data from a user terminal;
Learning the first log data to generate a location estimation model for the user using a naive Bayes classification algorithm; And
Analyzing second log data collected from the user terminal and estimating the current place of the user using the place estimating model,
The first log data includes at least one of time, place, and neighbor user information of the user terminal, and the second log data is a place using log data including at least one of time, neighbor user information of the user terminal. Estimation method. delete The method according to claim 6,
Grouping adjacent user information located within a preset range from the user terminal to generate a user group model using the naive Bayes classification algorithm and an EM clustering algorithm,
Estimating the current place of the user,
And a place estimating method using log data for estimating a place for the user according to the user group model using the place estimating model. The method according to claim 6,
Wherein the user terminal is a location estimation method using the log data for communicating data with another adjacent user terminal in at least one communication method of Bluetooth, Wi-Fi, 3G, 4G. delete

Images