JP7645852B2 - DATA ANALYSIS APPARATUS, DATA ANALYSIS SYSTEM, DATA ANALYSIS METHOD, AND COMPUTER PROGRAM - Google Patents

Description

The present invention relates to technology for analyzing user movement patterns.

There is known a technology that analyzes a user's movement patterns using observation values from sensors installed in advance at multiple locations. Here, "movement patterns" refers to information indicating when and where a user was (near which sensor), and movement patterns can be considered as a user's behavioral pattern. Such movement patterns can be used, for example, in planning measures such as behavior guidance within a facility. For example, Patent Document 1 describes a data analysis device that analyzes sensor data from sensors installed at multiple locations, and analyzes movement patterns (movement routes) by tracing routes with large amounts of movement between two locations based on received sensor data.

JP 2019-175240 A

Such technology generally determines that "the user is moving near a sensor with a strong signal reception strength." However, particularly inside a building, radio wave shielding or reflection by the building can occur, which can result in missing sensor observation data (in other words, when the signal from the user terminal cannot be received) or when observation data for a certain user terminal is acquired simultaneously by multiple sensors (in other words, when a signal from a certain user terminal is received simultaneously by multiple sensors). In such cases, there was a problem that analyzing a user's movement pattern solely from the signal reception strength by the sensors leads to a decrease in analysis accuracy and is not appropriate.

In this regard, the technology described in Patent Document 1 analyzes movement patterns by tracing routes with large amounts of movement between two points when observation data from a sensor is missing. However, there is an issue that routes with large amounts of movement between two points are simply "routes that many people have traveled," and therefore do not necessarily match the actual movement patterns of the target users. This issue is particularly pronounced when moving between two points in a small space such as inside a building, where many routes are possible.

The present invention has been made to solve at least some of the problems described above, and aims to improve the accuracy of analysis in technology that analyzes user movement patterns.

The present invention has been made to solve at least part of the above-mentioned problems, and can be realized in the following form: A data analysis device comprising: an acquisition unit that acquires observation data observed by sensors that are installed at multiple locations and receive signals transmitted from user terminals carried by users, the observation data including a user identifier included in the signal, a reception strength of the signal by the sensor, an observation time when the reception of the signal is observed by the sensor, and an identifier of the sensor; a travel time calculation unit that calculates a travel time between the sensors by the user from the multiple observation data acquired over time; a storage unit that pre-stores a score calculation formula that is an equation including a value according to a constraint condition of the travel time between the sensors by the user and the reception strength; and an analysis unit that analyzes a travel pattern indicating when and where the user was, using a score obtained by applying the multiple observation data acquired over time and the calculated travel time to the score calculation formula. The present invention can also be realized in the following form.

(1) According to one aspect of the present invention, a data analysis device is provided. The data analysis device includes an acquisition unit that acquires observation data observed by sensors installed at multiple locations, the observation data including a user identifier, a signal reception strength by the sensor, an observation time when the signal reception by the sensor was observed, and an identifier of the sensor, a travel time calculation unit that calculates the travel time between the sensors by the user from the multiple pieces of observation data acquired over time, and an analysis unit that analyzes a travel pattern that indicates when and where the user was by calculating a score using the multiple pieces of observation data acquired over time and the calculated travel time.

According to this configuration, the analysis unit calculates a movement pattern (information indicating when and where the user was) by calculating a score using the movement time between sensors by the user in addition to multiple observation data acquired over time. In this way, by taking the movement time between sensors by the user into consideration in the score calculation, it is possible to calculate a score based on the principles of human behavior, such as eliminating high-speed movement that exceeds the generally assumed movement speed of humans. As a result, it is possible to calculate a more reliable movement pattern compared to calculation of a movement pattern based on a route with a large amount of movement between two points (a route traveled by many people). In other words, according to this configuration, it is possible to improve the analytical accuracy in the technology for analyzing a user's movement pattern.

(2) In the data analysis device of the above form, the analysis unit may have a predetermined constraint condition for the travel time, and if, in the score calculation, the travel time between one of the sensors and another of the sensors does not satisfy the constraint condition, the travel between the one sensor and the other sensor may be excluded from the travel series.
According to this configuration, the analysis unit has a constraint condition for a predetermined movement time, and in score calculation, if the movement time between one sensor and another sensor does not satisfy the constraint condition, the movement between the one sensor and another sensor is excluded from the movement sequence. Therefore, high-speed movement that exceeds the generally assumed movement speed of a human can be excluded, and score calculation based on the principle of human behavior becomes possible.

(3) In the data analysis device of the above aspect, the analysis unit may use, as an index for calculating the score, an index for maximizing the sum of the reception strengths of the plurality of pieces of observation data acquired over time.
According to this configuration, the analysis unit uses an index for maximizing the sum of the reception strength among multiple observation data acquired over time as an index for calculating the score, making it possible to calculate the score based on the law that "the strength of the reception strength is proportional to the reliability of the observation data."

(4) In the data analysis device of the above form, the analysis unit may use an index for minimizing the amount of movement between the sensors by the user among the multiple pieces of observation data acquired over time as an index for calculating the score.
According to this configuration, the analysis unit uses an index for minimizing the amount of movement between sensors by the user from multiple observational data acquired over time as an index for calculating the score, making it possible to calculate a score based on the human behavior principle that "behavior involving less movement is more likely."

(5) In the data analysis device of the above form, the analysis unit may use, as an index for calculating the score, an index for maximizing the number of observations of a certain reception strength or greater among the multiple observation data acquired over time, and for minimizing the amount of movement between the sensors by the user.
According to this configuration, the analysis unit uses, as an index for calculating the score, an index for maximizing the number of observations with a certain reception strength or more among the multiple observation data acquired over time, and minimizing the amount of movement between sensors by the user. This makes it possible to calculate a score that satisfies both the law that "the strength of reception strength is proportional to the reliability of the observation data" and the human behavior principle that "behavior with less movement is more likely," and it is possible to further improve the accuracy of the analysis of the movement pattern.

(6) According to one aspect of the present invention, there is provided a data analysis system comprising: the data analysis device of the above aspect; and a data server configured to acquire and accumulate the observation data from each of the plurality of sensors over time.
According to this configuration, the data analysis device that analyzes the movement patterns and the data server that collects and stores the observation data are provided separately, so that the processing load can be distributed.

(7) According to one aspect of the present invention, there is provided a data analysis system comprising: the data analysis device of the above aspect; and a sensor that communicates with a terminal carried by the user to acquire an identifier of the user from the terminal.
According to this configuration, the sensor can acquire the user's identifier from the user terminal.

The present invention can be realized in various forms, such as a data analysis device that analyzes movement patterns, a server device that analyzes movement patterns and provides some kind of service, a method executed in an information processing device to realize the functions of each of these devices, a system including each of these devices, a computer program to realize the functions of each of these devices and systems, a server device for distributing the computer program, a non-transitory storage medium that stores the computer program, etc.

1 is a schematic diagram of a data analysis system according to an embodiment of the present invention. FIG. 11 is a sequence diagram illustrating an example of a procedure of a travel pattern analysis process. FIG. 13 is a diagram illustrating the analysis of travel patterns. FIG. 13 is a diagram illustrating the analysis of travel patterns. FIG. 1 is an explanatory diagram showing an example of a travel pattern. 11 is a diagram illustrating a process to be performed when observation data from a sensor is missing. FIG. FIG. 13 is a schematic diagram of a data analysis system according to a second embodiment. FIG. 11 is a sequence diagram illustrating an example of a procedure of a travel pattern analysis process according to the second embodiment.

First Embodiment
FIG. 1 is a schematic diagram of a data analysis system 1 according to an embodiment of the present invention. The data analysis system 1 is a system that analyzes and outputs the movement pattern of a user P. Here, the "movement pattern" is information indicating when and where the user P was (near which sensor). The movement pattern can be regarded as the behavior pattern of the user P. The data analysis system 1 includes a data analysis device 100, a data server 200, and a plurality of sensors S1 to S3. The number of sensors included in the data analysis system 1 is not limited to the illustrated example (three) and can be determined arbitrarily. In FIG. 1 and the following description, one user P is illustrated as an example, but the data analysis system 1 can analyze and output the movement patterns of each user by performing the processing described below in parallel for multiple users.

User P possesses a terminal (hereinafter referred to as "user terminal PD"). The user terminal PD is, for example, any device such as a smartphone, a wearable device, a game console, or a personal computer. The user terminal PD periodically transmits a signal including an identifier that can uniquely identify the user terminal PD (hereinafter also referred to as "identifier of the user terminal PD") by wireless communication. As the identifier of the user terminal PD, any identifier can be used, such as a MAC address (Media Access Control address) assigned in advance to a wireless communication device built into the terminal, or an ID assigned when an application is installed. Here, the identifier of the user terminal PD can be regarded as the identifier of the user P who possesses the user terminal PD (user identifier), and the location of the user terminal PD can be regarded as the location of the user P who possesses the user terminal PD. Note that a dedicated tag capable of transmitting an identifier may be used instead of the user terminal PD.

The sensors S1 to S3 communicate with the user terminal PD to obtain information from the user terminal PD. The sensors S1 to S3 may be sensors capable of receiving signals transmitted from the user terminal PD or access points. The sensors S1 to S3 are each assigned an identifier (hereinafter also referred to as a "sensor identifier") that can uniquely identify each sensor. The sensors S1 to S3 obtain the information shown in the following a1 to a3 by receiving a signal from the user terminal PD. The information a1 to a3 received by the sensors S1 to S3 and the identifier of the sensor shown in a4 are collectively referred to as "observation data."
(a1) User identifier: in other words, the identifier of the user terminal PD.
(a2) Reception strength: The reception strength of the signal from the user terminal PD by the sensor.
(a3) Observation time: the time when the sensor observes receipt of a signal from the user terminal PD. The observation time is, for example, a combination of date and time.
(a4) Sensor Identifier: An identifier assigned to the sensor that received the signal.

The data analysis device 100 is a server that analyzes and outputs the movement pattern of user P. The data analysis device 100 includes a CPU 110, a storage unit 120, a communication unit 130, and a memory 140, and each unit is connected to each other by a bus (not shown). The CPU 110 executes a computer program stored in the memory 140 to control each unit of the data analysis device 100, and also functions as an acquisition unit 111, a movement time calculation unit 112, and an analysis unit 113. The acquisition unit 111 acquires observation data (a1 to a4). The movement time calculation unit 112 calculates the movement time between sensors by user P from the observation data (a1 to a4). The analysis unit 113 analyzes the movement pattern of user P by calculating a score using the observation data (a1 to a4) and the calculated movement time. Details will be described later.

The storage unit 120 is composed of a hard disk, a flash memory, a memory card, etc. A score calculation formula (not shown, described in detail later) is pre-stored in the storage unit 120. The communication unit 130 controls communication with other devices via a communication interface (not shown). The other devices may include the data server 200, the user terminal PD, and other servers (not shown), etc.

The data server 200 is a server that acquires observation data (a1 to a4) over time from the sensors S1 to S3 and accumulates a bundle of observation data for each user P as an observation data set. The data server 200 comprises a CPU 210, a storage unit 220, a communication unit 230, and a memory 240, and each unit is connected to each other by a bus (not shown). The CPU 210 controls each unit of the data server 200 by executing a computer program stored in the memory 240, and also functions as a control unit 211. The control unit 211 acquires observation data (a1 to a4) over time from the sensors S1 to S3 and stores it in the storage unit 220. Details will be described later.

The storage unit 220 is composed of a hard disk, a flash memory, a memory card, etc. The storage unit 220 stores multiple pieces of observation data (a1 to a4) acquired over time by the control unit 211 from the sensors S1 to S3 through the movement pattern analysis process described below. Here, a bundle of multiple observation data (a1 to a4) in which the user identifier a1 is the same and any of the other information a2, a3, a4 differs is also called an observation data set 221. The communication unit 230 controls communication with other devices via a communication interface (not shown). The other devices may include the data analysis device 100, the sensors S1 to S3, the user terminal PD, and other servers (not shown), etc.

The sensors S1 to S3, the data analysis device 100, and the data server 200 are connected to each other via a network (e.g., the Internet) so that they can communicate with each other.

Figure 2 is a sequence diagram showing an example of the procedure for the travel pattern analysis process. For convenience of illustration, the user terminal PD and the sensor S3 described in Figure 1 are omitted in Figure 2.

In step S12, the sensor S1 receives a signal from the user terminal PD. In step S14, the sensor S1 transmits the user identifier a1, the signal reception strength a2, and the observation time a3 at which the signal reception was observed, contained in the received signal, as well as the sensor S1 identifier a4, to the data server 200 as "observation data". In step S16, the control unit 211 of the data server 200 stores the observation data in the storage unit 220. Steps S12 to S16 are collectively referred to as the reception and storage process (S10) by the sensor S1. Step S10 is executed each time the sensor S1 receives a signal.

In step S22, the sensor S2 receives a signal from the user terminal PD. In step S24, the sensor S2 transmits the user identifier a1, the signal reception strength a2, and the observation time a3 at which the signal reception was observed, contained in the received signal, as well as the sensor S2 identifier a4, to the data server 200 as "observation data". In step S26, the control unit 211 of the data server 200 stores the observation data in the storage unit 220. Steps S22 to S26 are collectively referred to as the reception and storage process (S20) by the sensor S2. Step S20 is executed each time the sensor S2 receives a signal.

The reception and storage process described in steps S10 and S20 is executed independently by each sensor included in the data analysis system 1.

In step S30, the data analysis device 100 determines whether or not there is a trigger to start the analysis of the travel pattern. The trigger can be determined arbitrarily, and can be, for example, the start of a specific application installed on the user terminal PD, the receipt of a request from the application that has already been started, a specific amount of observed data (a1 to a4) accumulated for user P, or a certain fixed timing. When a trigger is generated, the data analysis device 100 acquires the identifier of user P who is the subject of the analysis of the travel pattern. Note that when the analysis of the travel pattern is started at a certain fixed timing, the process described below can be performed for multiple users, not limited to a specific user P. For ease of explanation, the analysis of the travel pattern for a certain user P is described below.

In step S32, the acquisition unit 111 of the data analysis device 100 requests the observation dataset 221 from the data server 200. At this time, the acquisition unit 111 also transmits the identifier of user P acquired in step S30, thereby specifying the user P from whom the observation dataset 221 is to be acquired. In step S34, the control unit 211 of the data server 200 acquires the observation dataset 221 for the specified user P from the storage unit 220, and transmits it to the data analysis device 100. As described above, this observation dataset 221 includes multiple pieces of observation data (a1 to a4) acquired over time, in which the user identifier a1 is the same and any of the other information a2, a3, a4 is different.

Figures 3 and 4 are diagrams for explaining the analysis of the movement pattern. In Figures 3 and 4, the contents of the observation data set 221 for a certain user P are shown typically. In Figures 3 and 4, one observation data (a1 to a4) in the observation data set 221 is represented by one circle. The number in the circle represents the cumulative probability of the reception strength a2. The reception strength a2 is a value expressed in units such as dBm, but in this embodiment, the cumulative probability of the reception strength (a value from 0 to 1) is used as the reception strength a2. However, as the reception strength a2, instead of the cumulative probability of the reception strength, the value of the reception strength expressed in units such as dBm may be used as it is, or other information related to the reception strength may be used. Hereinafter, the cumulative probability of the reception strength is also simply referred to as "reception strength". The vertical axis position where the circle is placed corresponds to the identifier a4 of the sensor from which the observation data was acquired. The horizontal axis position where the circle is placed corresponds to the observation time a3 from which the observation data was acquired. In the example of FIG. 3(A), for example, at a certain time t1, for a certain user P, observation data with a reception strength of 0.1 is acquired by sensor S1, observation data with a reception strength of 0.4 is acquired by sensor S2, and observation data with a reception strength of 0.6 is acquired by sensor S3.

In step S36, the analysis unit 113 of the data analysis device 100 analyzes the movement pattern of the user P using the observation data set 221. Specifically, the analysis unit 113 calculates the first score S ^{q n} by applying each piece of observation data (a1 to a4) in the observation data set ₂₂₁ to the following formula (1). Details of each variable shown in formula (1) are as shown in the following b1 to b4.
(b1) k: A variable used for score calculation (incremented from 1 to n-1).
(b2) x _n , x _k : the nth observation sensor number, the kth observation sensor number.
(b3) Δx _k x _n : a value according to the constraint on the travel time between sensors x _k and x _n by user P, and a value determined by equation (2).
(b4) q _n : Reception strength a2.

To determine Δx _k x _n in formula (1), the travel time calculation unit 112 calculates the actual travel time taken by the user P between one sensor x _k and another sensor x _n . The actual travel time can be calculated from the absolute value of "observation time a3 of the observation data by sensor x _n - observation time a3 of the observation data by sensor x _k " (x _k -x _n in formula (2) above). The travel time constraint is stored in the storage unit 120 in advance, in which standard travel times according to the distance between each sensor are measured through experiments or the like. If the actual travel time between sensor x _k and sensor x _n is equal to or greater than the standard travel time between sensor x _k and sensor x _n in the storage unit 120, the analysis unit 113 determines that x _k -x _n satisfies the constraint condition, and sets Δx _k x _n = 1. On the other hand, if the actual travel time between sensor x _k and sensor x _n is shorter than the standard travel time between sensor x _k and sensor x _n in the storage unit 120, the analysis unit 113 determines that x _k -x _n does not satisfy the constraint condition, and sets Δx _k x _n =-∞. As is clear from the formula (1), if the actual travel time (x _k -x _n ) between sensor x _k and sensor x _n satisfies the constraint condition, Δx _k x _n =1 is set, and the movement between sensor x _k and sensor x _n is considered as something that can actually happen. On the other hand, if the actual travel time ₍ x _k -x _n ) between sensor x k and _sensor x n does not satisfy the constraint condition, Δx _k x _n =-∞ is set, and the movement (route) between sensor x _k and sensor x _n is excluded as something that cannot actually happen.

A specific example will be described with reference to FIG. 3 and FIG. 4. The analysis unit 113 substitutes the first observation data (n=1) into the formula (1) as shown in FIG. 3(A). As a result, the first score S ^q ₁ = the value of the reception strength a2 = 0.6. The analysis unit 113 substitutes the second observation data (n=2) into the formula (1) as shown in FIG. 3(B). As a result, the first score S ^q ₂ = the value of the reception strength a2 = 0.4. The travel time calculation unit 112 calculates the actual travel time between the sensor S3 and the sensor S2. As described above, the first observation data (n=1) by the sensor S3 and the second observation data (n=2) by the sensor S2 are both acquired at the time t1. Therefore, the absolute value of "observation time a3 of the observation data by the sensor S3 - observation time a3 of the observation data by the sensor S2" is close to 0. Therefore, the analysis unit 113 sets Δx _k x _n =-∞. As a result, as shown by the cross marks in Fig. 3B, the movement between the sensors S3 and S2 is excluded from the movement sequence to be considered.

As shown in FIG. 4(A), the analysis unit 113 excludes the third observation data (n=3) from consideration of the movement pattern. In this embodiment, the analysis unit 113 performs a process of excluding the observation data from consideration of the movement pattern when the value of the reception strength a2 is equal to or less than a predetermined reception strength threshold. This reception strength threshold is determined in advance and stored in the storage unit 120. Note that the analysis unit 113 may or may not perform the process described in FIG. 4(A).

The analysis unit 113 substitutes the fourth observation data (n=4) into the formula (1) as shown in FIG. 4B. When the user P moves from the sensor S3 to the sensor S2, the first score ^Sq4 = _0.6 +0.7=1.3. The travel time calculation unit 112 calculates the actual travel time between the sensor S3 and the sensor S2, and the analysis unit 113 _{sets Δxkxn} =1, and as a result, as shown in FIG. 4B, the travel between the sensor S3 and the sensor S2 is considered as a travel sequence. On the other hand, when the user P moves from the sensor S2 to the sensor S2, _the first score ^Sq4 = _0.4 + _0.7 =1.1. The travel time calculation unit 112 calculates the actual travel time between the sensor S2 and the sensor S2, and as a result, as shown in FIG. 4B, the travel between the sensor S2 and the sensor S2 is considered as a travel sequence.

The analysis unit 113 performs the above process on all data in the observation data set 221 to analyze one movement pattern with the largest first score S ^q _n . In this manner, in the first score S ^q _n , the sum of the reception strengths is maximized by using the variable q _n . In step S38 of FIG. 2, the analysis unit 113 of the data analysis device 100 outputs the movement pattern obtained by the analysis.

5 is an explanatory diagram showing an example of a movement sequence. As shown in FIG. 5, the movement sequence MS includes data in which a plurality of pairs of observation time and sensor identifiers are arranged in chronological order in addition to the identifier of the user P. The pair of observation time and sensor identifier is information indicating when and where the user P was (near which sensor). Note that the form of the movement sequence MS is merely an example, and instead of the sensor identifier, sensor position information derived from the sensor identifier may be used. The output form of the movement sequence MS can also be determined arbitrarily. For example, the movement sequence MS may be output by transmitting it from the data analysis device 100 to the user terminal PD (or another server that provides some kind of service using the movement sequence MS), or may be output by saving the movement sequence MS in the storage unit 220 of the data analysis device 100, or may be output by displaying the movement sequence MS on the display unit of the data analysis device 100. Note that step S38 of outputting the movement sequence may be omitted.

2, the analysis unit 113 of the data analysis device 100 may calculate the second score S d n using the following formula (3) instead of formula (1) that maximizes the sum of the reception intensities. In the second score S ^d _n , the amount of movement between sensors by the user P is minimized by using the variable tx _k x _n . Details of each variable shown in formula (3) are as shown in ^the following c1 _and c2 in addition to the above-mentioned b1 to b3.
(c1) α: an arbitrary coefficient representing a weight for adding an observation. By setting the weight coefficient α to be larger than the actual travel time between the sensor x _k and the sensor x _n , it is possible to extract a travel sequence that obtains as many observations as possible.
(c2) tx _k x _n : the amount of movement between sensor x _k and sensor x _n (in other words, the movement distance). The amount of movement (movement distance) between each sensor is measured in advance and stored in the storage unit 120.

2, the analysis unit 113 of the data analysis device 100 may calculate the third score S d ' n using the following formula (4) instead of formula (1) that maximizes the sum of the reception strengths. In the third score S ^{d '} _n , the variable β·I[r _n >K] and the variable _{tx k} x _n are used to maximize the number of observations with a certain reception strength or more and minimize the amount of movement between sensors. Details of each variable shown in formula (4) are as shown in the following d1 to d4 in addition to the above-mentioned b1 to b3, and c1 and c2.
(d1) β: an arbitrary weighting factor to be given to observed values equal to or greater than K dBm. The weighting factor β is set to a value sufficiently larger than the weighting factor α (β>>α).
(d2) I[·]: indicator function. If the value in the parentheses is true, it takes the value 1, and if the value is false, it takes the value 0. Here, "I" is a substitute character for the "symbol expressing the Arabic numeral 2 in Roman numerals" that appears in formula (4).
(d3) r _n : Reception strength a2 of sensor x _n . Note that for variable d3, the reception strength value itself expressed in units such as dBm is used, rather than the cumulative probability of the reception strength.
(d4) K: an arbitrary constant for determining whether or not to take into account the weighting coefficient β. In other words, the threshold value of the reception strength to be taken into account in the score.

Figure 6 is a diagram for explaining the process when the observation data by the sensor is missing. Figure 6 shows the contents of the observation data set 221 when the observation data (a1 to a4) by the sensors S1 to S3 are missing, in other words, when the sensors S1 to S3 cannot receive a signal from the user terminal PD for a certain period tx. In such a case, the analysis unit 113 can perform the above-mentioned process by assuming that the observation data was acquired by the same sensor (sensor S3 in the illustrated example) that received the signal just before the period tx ends until just before the period tx ends. In this case, the reception strength a2 may be the same as the reception strength a2 of the immediately preceding observation data (0.3 in the illustrated example), or may be an arbitrary default value. This is because the signal reception by the sensors S1 to S3 is easily detected at the timing of handover (in other words, when the sensors S1, S2, and S3 are switched). In other words, the absence of signal reception by the sensors S1 to S3 is because there is a high possibility that there is no handover (in other words, the user P remains in the same position). The observation data complementation process described in FIG. 6 can be performed, for example, as preprocessing for the analysis of movement patterns (step S36).

As described above, according to the data analysis device 100 of the first embodiment, the analysis unit 113 calculates the movement sequence MS (information indicating when and where the user was) by calculating the score (calculation of S ^q _n , S ^d _n , S ^d ′ _n ) using the movement time between sensors by the user P in addition to the multiple observation data (observation data set 221) acquired over time. In this way, as described in formula (2), in the score calculation, by considering the movement time between sensors by the user P, it is possible to calculate the score based on the principle of human behavior, such as excluding high-speed movement that exceeds the generally assumed human movement speed (when Δx _k x _n =−∞). As a result, it is possible to calculate a more likely movement sequence MS compared to the calculation of a movement sequence MS based on a route with a large amount of movement between two points (a route on which many people have traveled). That is, according to the data analysis device 100 of the present embodiment, the analysis accuracy can be improved in the technology for analyzing the movement sequence of the user.

In addition, according to the data analysis device 100 of the first embodiment, the analysis unit has a predetermined constraint condition for travel time, and in score calculation, if the travel time between one sensor and another sensor does not satisfy the constraint condition, the travel between the one sensor and the other sensor is excluded from the travel sequence. This makes it possible to exclude high-speed travel that exceeds the generally assumed travel speed of humans, and makes it possible to calculate the score based on the principles of human behavior.

Furthermore, in the data analysis device 100 of the first embodiment, according to formula (1), the analysis unit 113 uses the first score S ^q _n , which is an index for maximizing the sum of reception strengths among multiple observation data (observation data set 221) acquired over time, as an index for score calculation, so that it becomes possible to calculate the score based on the law that "the strength of reception strength is proportional to the reliability of observation data."

Furthermore, in the data analysis device 100 of the first embodiment, according to formula (3), the analysis unit 113 uses the second score S ^d n, which is an index for minimizing the amount of movement between sensors by the user from among multiple observation data (observation data set 221) acquired over time, as an index for _score calculation, so that it is possible to calculate the score based on the human behavior principle that "behavior with less movement is more likely."

Furthermore, in the data analysis device 100 of the first embodiment, according to formula (4), the analysis unit 113 uses, as an index for score calculation, the third score Sd'n, which is an index for maximizing the number of observations with a certain reception strength or more among a plurality of observation data (observation data ^set ₂₂₁ ) acquired over time and minimizing the amount of movement between sensors by the user. This makes it possible to calculate a score that satisfies both the law that "the strength of reception strength is proportional to the reliability of observation data" and the human behavior principle that "behavior with less movement is more likely," and it is possible to further improve the analysis accuracy of the movement pattern.

Furthermore, according to the data analysis system 1 of the first embodiment, the data analysis device 100 that analyzes the movement sequence MS and the data server 200 that collects and stores the observation data (a1 to a4) are separately provided, so that the processing load can be distributed. In addition, the observation data (a1 to a4) can be acquired from the user terminal PD by the sensors S1 to S3.

Second Embodiment
7 is a schematic diagram of a data analysis system 1A according to the second embodiment. In the second embodiment, the data analysis system 1A is configured without using the data server 200. The data analysis system 1A according to the second embodiment includes a data analysis device 100A instead of the data analysis device 100 in the configuration described in the first embodiment, and does not include the data server 200.

The data analysis device 100A includes an acquisition unit 111A instead of the acquisition unit 111, and a storage unit 120A instead of the storage unit 120. The acquisition unit 111A differs from the first embodiment described in FIG. 2 in the content of the movement pattern analysis process. The storage unit 120A stores a plurality of pieces of observation data (a1 to a4) acquired over time from the sensors S1 to S3 by the acquisition unit 111A through the movement pattern analysis process described below.

Figure 8 is a sequence diagram showing an example of the procedure of the movement pattern analysis process of the second embodiment. Hereinafter, only the parts different from the first embodiment described in Figure 2 will be described. In step S14A, the sensor S1 transmits the observation data (a1 to a4) to the data analysis device 100A. In step S16A, the acquisition unit 111A of the data analysis device 100A stores the observation data in the storage unit 120A. In step S24A, the sensor S2 transmits the observation data (a1 to a4) to the data analysis device 100A. In step S26A, the acquisition unit 111A of the data analysis device 100A stores the observation data in the storage unit 120A. In step S40, the acquisition unit 111A of the data analysis device 100A acquires the observation data set 121 for the user P from the storage unit 120A. The subsequent processes are the same as those of the first embodiment described in Figure 2.

In this way, the configuration of the data analysis system 1A can be modified in various ways, and the system may be configured without using the data server 200. In addition, the functional parts of the data analysis device 100 described in the first embodiment and the data analysis device 100A described in the second embodiment may be realized by multiple server devices working together. In this way, the data analysis system 1A can be configured flexibly, improving the scalability of the system.

<Modifications of this embodiment>
The present invention is not limited to the above-described embodiment, and can be embodied in various forms without departing from the spirit of the present invention. For example, a part of the configuration realized by hardware may be replaced by software, and conversely, a part of the configuration realized by software may be replaced by hardware. In addition, for example, the following modifications are also possible.

[Modification 1]
In the above embodiment, the configurations of the data analysis systems 1 and 1A are exemplified. However, as described in the second embodiment, the configuration of the data analysis system 1 may adopt any aspect. For example, the data analysis system 1 may further include a service server that provides some kind of service by using the outputted mobility pattern MS. For example, the sensors S1 to S3 of the data analysis system 1 may be realized by different types of sensors.

[Modification 2]
In the above embodiment, the travel pattern analysis process (FIGS. 2 and 8) has been described by way of an example of a processing procedure. However, these processing procedures can be modified in various ways, and the processing contents in each step can be added/omitted/modified, and the execution order of the steps (procedures) can be changed.

For example, the information a1 to a4 exemplified as the observation data are merely examples, and some of the information may be omitted, or other information may be included. For example, when the analysis unit 113 calculates a movement pattern using the second score S ^d _n , the reception strength a2 of the observation data may be omitted. For example, the observation data may further include some parameter used for score calculation (calculation of S ^q _n , S ^d _n , and S ^d ' _n ). For example, the some parameter may be information indicating a communication method between the user terminal PD and the sensors S1 to S3.

[Modification 3]
It is possible to appropriately combine the features of the above-mentioned first and second embodiments and modifications 1 and 2. For example, the features described in modifications 1 and 2 may be adopted in the configuration of the second embodiment (a configuration without using the data server 200).

Although this aspect has been described above based on the embodiment and modified examples, the embodiment of the above-mentioned aspect is intended to facilitate understanding of this aspect and does not limit this aspect. This aspect may be modified or improved without departing from the spirit and scope of the claims, and equivalents are included in this aspect. Furthermore, if a technical feature is not described as essential in this specification, it may be deleted as appropriate.

The present invention can also be realized in the following forms.
[Application Example 1]
1. A data analysis device, comprising:
an acquisition unit that acquires observation data observed by sensors installed at a plurality of locations, the observation data including a user identifier, a signal reception strength by the sensor, an observation time when the signal reception was observed by the sensor, and an identifier of the sensor;
a travel time calculation unit that calculates a travel time between the sensors by the user from a plurality of pieces of observation data acquired over time;
an analysis unit that analyzes a travel pattern that indicates when and where the user was by calculating a score using a plurality of pieces of observation data acquired over time and the calculated travel time;
A data analysis device comprising:
[Application Example 2]
The data analysis device according to Application Example 1,
The analysis unit has a predetermined constraint condition for the travel time, and if, in the score calculation, the travel time between one of the sensors and another of the sensors does not satisfy the constraint condition, the travel between the one of the sensors and the another of the sensors is excluded from the travel series.
[Application Example 3]
The data analysis device according to the first or second aspect of the present invention,
The analysis unit uses an index for maximizing the sum of the reception strengths of the multiple pieces of observation data acquired over time as an index for calculating the score.
[Application Example 4]
The data analysis device according to any one of Application Examples 1 to 3,
The analysis unit uses an index for minimizing the amount of movement between the sensors by the user among the multiple pieces of observation data acquired over time as an index for calculating the score.
[Application Example 5]
The data analysis device according to any one of Application Examples 1 to 4,
The analysis unit uses an index for calculating the score that maximizes the number of observations of a certain reception strength or greater among the multiple observation data acquired over time, and minimizes the amount of movement between the sensors by the user.
[Application Example 6]
1. A data analysis system, comprising:
The data analysis device according to any one of Application Examples 1 to 5,
a data server that acquires and accumulates the observation data from each of the plurality of sensors over time;
A data analysis system comprising:
[Application Example 7]
1. A data analysis system, comprising:
The data analysis device according to any one of Application Examples 1 to 5,
A data analysis system comprising a sensor that communicates with a terminal carried by the user to obtain an identifier of the user from the terminal.
[Application Example 8]
A data analysis method, comprising:
acquiring observation data observed by sensors installed at a plurality of locations, the observation data including a user identifier, a signal reception strength by the sensor, an observation time when the signal reception was observed by the sensor, and an identifier of the sensor;
calculating a travel time between the sensors by the user from a plurality of pieces of observation data acquired over time;
analyzing a travel pattern indicating when and where the user was by calculating a score using the plurality of pieces of observation data acquired over time and the calculated travel time;
Perform the data analysis method.
[Application Example 9]
A computer program comprising:
A function of acquiring observation data observed by sensors installed at a plurality of locations, the observation data including a user identifier, a signal reception strength by the sensor, an observation time when the signal reception was observed by the sensor, and an identifier of the sensor;
A function of calculating a travel time between the sensors by the user from a plurality of pieces of the observation data acquired over time;
a function of analyzing a movement pattern that indicates when and where the user was by calculating a score using a plurality of pieces of the observation data acquired over time and the calculated movement time;
A computer program that achieves this.

1, 1A... Data analysis system 100, 100A... Data analysis device 110... CPU
111, 111A... Acquisition unit 112... Travel time calculation unit 113... Analysis unit 120, 120A... Storage unit 121... Observation data set 130... Communication unit 140... Memory 200... Data server 210... CPU
211: control unit 220: storage unit 221: observation data set 230: communication unit 240: memory S1 to S3: sensors

Claims (8)

1. A data analysis device, comprising:
an acquisition unit that acquires observation data observed by sensors that are installed at a plurality of locations and receive signals transmitted from user terminals carried by users , the observation data including a user identifier included in the signal , a reception strength of the signal by the sensor, an observation time at which the reception of the signal was observed by the sensor, and an identifier of the sensor;
a travel time calculation unit that calculates a travel time between the sensors by the user from a plurality of pieces of observation data acquired over time;
a storage unit that stores in advance a score calculation formula that is an equation including a value according to a constraint condition of a movement time between the sensors by the user and the reception strength;
an analysis unit that analyzes a movement pattern indicating when and where the user was, using a score obtained by applying the plurality of pieces of observation data acquired over time and the calculated movement time to the score calculation formula ;
A data analysis device comprising: The data analysis device according to claim 1 ,
The analysis unit has a predetermined constraint condition for the travel time, and if, in the score calculation, the travel time between one of the sensors and another of the sensors does not satisfy the constraint condition, the travel between the one of the sensors and the another of the sensors is excluded from the travel series. The data analysis device according to claim 2,
The analysis unit analyzes the travel pattern by using the score calculation formula defined by the following formula (1),

In the above formula (1),
The left side of the equal sign is the score,
on the right hand side of the equal sign, k is a variable, x _n is an identifier of the sensor in the n-th observation data, x _k is an identifier of the sensor in the k-th observation data, Δx _k x _n is a value according to a constraint on the travel time by the user between sensor x _k and sensor x _n and is a value determined by the above formula (2), and q _n is the reception strength in the n-th observation data . The data analysis device according to claim 2,
The storage unit stores in advance the amount of movement between the sensors,
The analysis unit analyzes the travel pattern by using the score calculation formula defined by the following formula (4),

In the above formula (4),
The left side of the equal sign is the score,
a data analysis device in which, on the right side of the equal sign, k is a variable, x _n is an identifier of the sensor in the n-th observation data, x _k is an identifier of the sensor in the k-th observation data, Δx _k x _n is a value according to a constraint condition of a movement time between sensor x _k and sensor x _n by the user and is a value determined by the above formula (2), α is a weighting coefficient, tx _k x _n is the movement amount between sensor x _k and sensor x _n , β is a weighting coefficient that takes a value sufficiently larger than the weighting coefficient α, I[·] is an indicator function that takes 1 when the value in parentheses is true and 0 when it is false, r _n is the reception strength of sensor x _n , and K is a constant for determining whether to take the weighting coefficient β into consideration or not . 1. A data analysis system, comprising:
A data analysis device according to any one of claims 1 to 4 ,
a data server that acquires and accumulates the observation data from each of the plurality of sensors over time;
A data analysis system comprising: 1. A data analysis system, comprising:
A data analysis device according to any one of claims 1 to 4 ,
The sensor . A data analysis method, comprising :
acquiring observation data observed by sensors installed at a plurality of locations and receiving signals transmitted from user terminals carried by users , the observation data including a user identifier included in the signal , a reception strength of the signal by the sensor, an observation time at which the reception of the signal was observed by the sensor, and an identifier of the sensor;
calculating a travel time between the sensors by the user from a plurality of pieces of observation data acquired over time;
reading out a score calculation formula stored in advance in the storage unit, the score calculation formula being an equation including a value according to a constraint condition of a movement time between the sensors by the user and the reception strength;
analyzing a movement pattern indicating when and where the user was, using a score obtained by applying the plurality of pieces of observation data acquired over time and the calculated movement time to the score calculation formula ;
Perform the data analysis method. A computer program comprising :
a function of acquiring observation data observed by sensors installed at a plurality of locations and receiving signals transmitted from user terminals carried by users , the observation data including a user identifier included in the signal , a reception strength of the signal by the sensor, an observation time at which the reception of the signal was observed by the sensor, and an identifier of the sensor;
A function of calculating a travel time between the sensors by the user from a plurality of pieces of the observation data acquired over time;
a function of reading out a score calculation formula stored in advance in the storage unit, the score calculation formula being an equation including a value according to a constraint condition of a movement time between the sensors by the user and the reception strength;
a function of analyzing a movement pattern indicating when and where the user was, using a score obtained by applying the plurality of pieces of observation data acquired over time and the calculated movement time to the score calculation formula ; and
A computer program that achieves this.

Images