JP4680804B2 - Information collection control device and information collection control method - Google Patents

Description

  The present invention relates to an information collection control device and an information collection control method.

  2. Description of the Related Art Conventionally, there has been proposed a technique for controlling an information collection operation for the purpose of improving information collection efficiency when collecting information on an external target using a plurality of information acquisition devices (see, for example, Non-Patent Document 1). ).

As a technique for improving the efficiency of information collection, for example, continuous information such as the position of a certain object is periodically measured (acquired) using an information acquisition device such as GPS (Global Positioning System). In a system that collects using the wireless band and reproduces the above continuous values by calculation from a part of the collected information, the difference between the true value of the object and the value calculated by the system is a certain threshold value. Techniques aimed at reducing the number of times of information acquisition and information collection in an information acquisition device while maintaining the following have been proposed. In this technique, a model that predicts a future value based on values acquired by the system from the past to the present is used, and a value at the time of next information collection is predicted by the model. Then, the predicted value is compared with the actually obtained value to obtain the divergence between the two. When the divergence is large, the frequency of acquisition and collection is high, and when the divergence is small, the acquisition and collection are performed. By reducing the frequency, the number of times of acquisition and collection is reduced as compared with the case where acquisition and collection are always performed at the maximum possible frequency by the information acquisition apparatus. This aims to achieve the purpose of efficient information acquisition and collection.
Ankur Jain and Edward Y. Chang: "Adaptive sampling for Sensor Networks," in Proc. International workshop on Data Management for sensor networks (2004).

  However, the information collection from the information acquisition device, depending on the use of the collected information, does not necessarily reduce the difference between the true value of the measurement target and the value obtained by the calculation as in the above example. There is no need to keep. For example, in a service providing system that determines the operation of each service based on information collected from an external information acquisition device and a plurality of predetermined service execution conditions and collects the determined operation, the information is collected from the information acquisition device. When the importance of this is low (for example, when information having a small influence on the operation of each service execution condition or the number of services to be referred to is small), no problem occurs even when the above-described divergence is large. In such a case, it is not necessary to keep the deviation between the true value of the measurement target and the value obtained by the calculation below a certain value. Therefore, when the importance level is low as described above, information collection and acquisition are not performed, so that information acquisition and service in the information acquisition apparatus can be maintained while keeping the service operation sequence (hereinafter referred to as “service operation”) equal. There was room for further reduction in the number of collections by the providing system (hereinafter referred to as “information collection times”).

  Therefore, the present invention provides an information collection control device and an information collection control method capable of obtaining information in the information acquisition device and reducing the number of information collections by the service providing system while keeping the operation of each service equal. With the goal.

  In order to achieve the above object, according to a first aspect of the present invention, a service is provided at a time when a service execution condition is satisfied based on information collected from one or more information acquisition devices and one or more service execution conditions. An information collection control device arranged in a service providing system to provide, (a) a prediction means for predicting information collected from the information acquisition device at a future time based on information collected from the information acquisition device in the past (B) When information is collected from one or more information acquisition devices used in the service execution condition based on the analysis result of the service execution condition using the prediction result by the prediction unit, the service execution condition is An execution condition satisfaction evaluation value deriving means for deriving an evaluation value based on a probability of being satisfied; and (c) comparing the evaluation value with a predetermined threshold value, and evaluating the evaluation value from the threshold value A threshold value comparison unit that identifies an information acquisition device that increases, and (d) an information acquisition device that has an evaluation value that is higher than the threshold value at the nearest future time among the information acquisition devices identified by the threshold value comparison unit. The gist of the present invention is an information collection control device including an information collection control means for transmitting an information collection instruction.

  According to the information collection control device according to the first feature, based on the information collected from the information acquisition device in the past, the information collected from the information acquisition device at this time is information only at a high probability that the service execution condition is satisfied. By collecting the information, it is possible to collect information only when the degree of importance is high, and it is possible to reduce the number of times of information collection and the number of times of information acquisition of the information acquisition device while keeping the operation of each service equal.

  Further, in the information collection control device according to the first feature, the service execution condition is described in the form of a Bayes decision tree, and the execution condition satisfaction evaluation value deriving means is a node in the Bayes decision tree based on a prediction result by the prediction means. Update the prior probabilities and set the evidence to the node corresponding to the information collection control device that derives the evaluation value using each candidate value that can be collected from the information collection control device. When the candidate value is set as evidence, a value with a high probability that the service execution condition is satisfied is checked based on the updated probability distribution of the node, and the value is collected from the information collection control device that derives the evaluation value The sum of probabilities may be used as the evaluation value.

  According to this information collection control device, when analyzing a service execution condition described in the form of a Bayesian decision tree, the execution condition satisfaction evaluation value can be immediately obtained by analyzing the Bayesian decision tree.

  A second feature of the present invention is arranged in a service providing system that provides a service at a time when the service execution condition is satisfied based on information collected from one or more information acquisition devices and one or more service execution conditions. And (a) predicting information collected from the information acquisition device at a future time based on information collected from the information acquisition device in the past, and (b) predicting step An evaluation value based on the probability that the service execution condition is satisfied when information is collected from one or more information acquisition devices used in the service execution condition based on the analysis result of the service execution condition using the prediction result of And (c) comparing the evaluation value with a predetermined threshold value to identify an information acquisition device whose evaluation value is higher than the threshold value. And (d) a step of transmitting an information collection instruction at the time to the information acquisition device whose evaluation value is higher than the threshold value at the nearest future time among the specified information acquisition devices. The gist is that it is a collection control method.

  In the information collection control method according to the second feature, the service execution condition is described in the form of a Bayesian decision tree, and in the step of deriving, the prior probabilities of the nodes in the Bayesian decision tree are calculated based on the prediction results by the prediction means. Update and set evidence for the node using each candidate value that can be collected from the information collection control device in the node corresponding to the information collection control device that derives the evaluation value, and set the candidate value among the previous candidate values When set as evidence, a value with a high probability that the service execution condition is satisfied is checked based on the updated probability distribution of the node, and the sum of the probability that the value is collected from the information collection control device that derives the evaluation value It may be an evaluation value.

  According to the present invention, it is possible to provide an information collection control device and an information collection control method capable of obtaining information in the information acquisition device and reducing the number of times of information collection by the service providing system while keeping the operation of each service equal. it can.

  Next, embodiments of the present invention will be described with reference to the drawings. In the following description of the drawings, the same or similar parts are denoted by the same or similar reference numerals. However, it should be noted that the drawings are schematic.

(Service provision system)
The service providing system according to the present embodiment provides a service at a time when the service execution condition is satisfied based on information collected from one or more information acquisition devices and one or more service execution conditions.

  As shown in FIG. 1, the service providing system according to the present embodiment includes a plurality of information acquisition devices 50 that acquire information, and a server 100 that receives information from the information acquisition device 50. Examples of the information acquisition device 50 include a terminal device having a GPS (Global Positioning System) that acquires position information, a temperature sensor, a humidity sensor, or a site on the Internet where weather forecast information or stock price information is registered. Can be mentioned.

  Further, the information acquisition device 50 and the server 100 communicate via the communication network 30. Examples of the communication network 30 include the Internet, and a LAN, WAN, intranet, or the like may be used.

  The server 100 includes a CPU 101 including a prediction unit, an execution condition satisfaction evaluation value deriving unit, a threshold comparison unit, and an information collection control unit, which will be described in detail later, an input unit 102, an output unit 103, a data holding unit 104, and a program holding unit 105. Is provided.

  The input unit 102 refers to a device such as a keyboard and a mouse. When an input operation is performed from the input unit 102, corresponding key information is transmitted to the processing control device (CPU) 101. The output unit 103 indicates a screen such as a monitor, and a liquid crystal display (LCD), a light emitting diode (LED) panel, an electroluminescence (EL) panel, or the like can be used.

  The data holding unit 104 holds service execution conditions, received data, and data being processed. The data holding unit 104 may use an internal storage device such as a RAM or an external storage device such as a hard disk or a flexible disk.

  The program storage unit 105 stores a control program for causing the processing control device (CPU) 101 to execute a prediction process, an execution condition satisfaction evaluation value derivation process, a threshold comparison process, an information collection control process, and the like described in detail later. It is. Examples of the recording medium include a RAM, a hard disk, a flexible disk, a compact disk, an IC chip, and a cassette tape. According to the recording medium holding such a program, the program can be easily stored, transported, sold, and the like.

  FIG. 2 schematically shows a configuration example of the server 100 and various information acquisition devices 50a, 50b,..., 50n (hereinafter collectively referred to as “information acquisition device 50”).

  The server 100 includes an information collection control device 10 and a service execution condition storage unit 20.

  The service execution condition storage unit 20 stores a plurality of service execution conditions 21a, 21b,..., 21n (hereinafter collectively referred to as “service execution conditions 21”) written in a format such as a Bayes decision tree. In the following, a case where there are a plurality of service execution conditions 21 will be described. However, there may be only one service execution condition 21. An example of the service execution condition 21 will be described later with reference to FIGS.

  The server 100 refers to information collected by communication with the information acquisition device 50 and executes a service according to the collected information and the service execution condition 21.

  Here, before describing the processing operation of each unit of the information acquisition apparatus 50, a method for determining a service operation in the server 100, which is a premise of the present invention, will be described. Here, it is assumed that the service execution condition 21 is described in the form of a Bayes decision tree. Examples of service execution conditions 21 described by a Bayesian decision tree are shown in FIGS.

  In the following, service operation determination using a Bayesian decision tree will be described taking the service execution condition (1) for the umbrella preparation recommendation service shown in FIG. 3 as an example.

  In this service execution condition (1), a probability relating to whether or not the current weather is rain is inferred using a state whether or not the humidity is high and a state whether or not the weather forecast is rainy. Then, based on the probability obtained by the inference, it is determined whether or not a message for recommending umbrella preparation should be displayed on the user terminal. Such a determination is made periodically without any explicit instruction from the user.

  Each node indicated by an ellipse in FIG. 3A represents a state variable used for determining the operation of the service. For example, a node described as high humidity (H) represents a state variable expressing whether or not the humidity is high, and the state variable takes two values, high (True) and low (False). The arrow between the nodes indicates that the state represented by the parent node (the original node from which the arrow appears) directly affects the state represented by the child node. In the service execution condition (1), the state of whether or not the humidity is high has a direct influence on the state of rain.

  Such an influence between nodes is expressed in the form of a conditional probability table. The conditional probability table is described for each node, and represents a probability distribution of values taken by the state of the node relative to the state of the parent node. For example, in the service execution condition (1), the value that the state of raining (R) takes is affected by the values of its parent nodes H (high humidity) and W (weather forecast is rainy). Each row in the probability table shows the probability that R (rain falls) will be true for each set of values taken by H (high humidity) and W (weather forecast is rain) (P (R | H, W)) Represents. For example, for a pair where H (high humidity) is True and W (weather forecast is rain) is True, the probability (P (R | H, W)) that R (rains) is True is 0.95. It is expressed.

  On the other hand, a predetermined probability (prior probability) is given to a node that does not have a parent (that is, a node that does not have an incoming arrow). For example, in the service execution condition (1), the probability of high humidity P (H) = 0.2 is expressed.

  By using the conditional probability table and the value of the prior probability, it is possible to calculate probabilities that other states can take when evidence for some states is obtained. For example, the probability of raining and the probability of not falling (P (R | H)) can be calculated when evidence is given for a condition of high humidity (H = True). Such a probability can be obtained from a value inside the Bayesian decision tree by a known calculation algorithm.

  A rectangle in the service execution condition represents a decision-making option in the decision tree. The rectangle (that is, operation (A)) in the service execution condition (1) indicates two operations that are recommended to carry an umbrella (a1) and do nothing (a2). Moreover, the rhombus in the service execution condition indicates the utility value in the state after the actual operation to be performed is determined. The utility value here is expressed in a table format as shown in FIG. 3B, and each row is the selected action and the value of the state variable that affects the utility at that time (the value of the state variable of the parent node of the utility). The utility value for each set is shown.

In service operation judgment using a Bayesian decision tree, the utility expected when EU inputs each node's evidence using currently available information and performs each operation with that evidence obtained (EU) Calculate Then, the operation having the largest utility (EU) obtained by the calculation is executed. For example, in the service execution condition (1), the expected utility before the information is obtained is obtained using the following equation (1).

  Here, ri represents each possible value of the state variable that rains. The expected utility of the action (a1) that recommends taking out an umbrella and the action (a2) that does nothing is calculated as follows.

EU (a1) = 0.2134 × 10 + 0.7866 × 6 = 6.8536
EU (a2) = 0.2134 × (-7) + 0.7866 × 9 = 5.5856
“0.2134” here is a group in which H (high humidity) is True and W (weather forecast is rain) is True, a group in which H is True and W is False, a group in which H is False and W is True, and It is obtained by the sum of the probabilities that R (raining) is true for each pair of H = False and W = False. That is, 0.2 × 0.1 × 0.95 + 0.2 × 0.9 × 0.6 + 0.8 × 0.1 × 0.9 + 0.8 × 0.9 × 0.02 = 0.2134. On the other hand, “0.7866” is obtained as 1−0.2134.

  In the above case, the expected utility EU (a1) of the action (a1) that recommends taking out the umbrella is larger than the expected utility EU (a2) of the action (a2) that does nothing, so the action that recommends taking out the umbrella I do. As such an operation, for example, information (such as a message recommending to take out an umbrella) is transmitted to the terminal device to perform screen display, voice notification, or the like. Along with this, a transmission unit for transmitting information to the terminal device is required in the server 100.

  The service described by each service execution condition obtains evidence regarding some nodes of the service execution condition from the information acquisition device 50. For example, the information acquisition device 50 used in the service execution condition (1) is a humidity sensor included in the user terminal and a weather forecast site on the Internet.

  Another example is the automatic manner mode service shown in FIG. In this service execution condition (2), the probability regarding whether or not the user is in a meeting is inferred using the state whether or not the user is in the conference room and the state whether or not the conference is reserved. Then, based on the probability obtained by inference, it is determined whether or not the user terminal should be set to the manner mode. The service described by each service execution condition obtains evidence regarding some nodes of the service execution condition from the information acquisition device 50. For example, the information acquisition device 50 used in the service execution condition (2) is an in-house system having reservation information and an indoor position measurement device.

  Furthermore, as another example, there is a service end notification service shown in FIG. In this service execution condition (3), the probability regarding whether or not the user leaves the seat and is dealing with the business is inferred using the state of whether or not he / she is at his / her own seat. Then, based on the probability obtained by the inference, it is determined whether or not the desired user should be notified of the end of the business (the user has returned to his / her seat). The service described by each service execution condition obtains evidence regarding some nodes of the service execution condition from the information acquisition device 50. For example, the information acquisition device 50 used in the service execution condition (3) is an indoor position measurement device.

  Subsequently, the operation of each unit of the information collection control device 10 will be described in more detail.

  The information collection control device 10 includes a prediction unit 11 (corresponding to a prediction unit according to the present invention), an execution condition satisfaction evaluation value deriving unit 12 (corresponding to an execution condition satisfaction evaluation value deriving unit according to the present invention), a threshold comparison unit 13 ( And an information collection control unit 14 (corresponding to the information collection control means according to the present invention).

  The prediction unit 11 predicts future time information based on information collected in the past from the information acquisition device 50. Here, a case where position information in a predefined two-dimensional coordinate system is acquired by using position positioning means such as GPS provided in a portable terminal carried by a certain user will be described as an example. For example, it is assumed that the user exists at the coordinates [20 m, 30 m] at the previous positioning and exists at the coordinates [20 m, 40 m] at the previous positioning. At this time, the movement speed can be calculated from the coordinates at the time of the previous and previous measurement, and the position at a certain time can be predicted from the previous position. By this prediction, the probability distribution of the user's position is obtained such that the probability of being within a circle of 5 m radius of [20 m, 50 m] at a future time t is 0.8, and the probability of being other than that is 0.2. Such prediction is performed using a known prediction technique.

  Based on the service execution condition 21 stored in the service execution condition storage unit 20 and the analysis result of the service execution condition using the prediction result by the prediction unit 11, the execution condition satisfaction evaluation value deriving unit 12 When information is collected from one or more information acquisition devices to be used, an evaluation value based on the probability that the service execution condition is satisfied is derived.

  Specifically, the execution condition satisfaction evaluation value deriving unit 12 inputs a probability distribution of values at a future time from the prediction unit 11 using information collected at the time of past determination. The execution condition satisfaction evaluation value deriving unit 12 updates the distribution of prior probabilities for possible values of each node according to the result. For example, because the previous weather forecast was clear, the probability that the forecast at time τ was rain was 0.1, and the humidity at the time of previous information collection was low and the humidity did not tend to rise, so the probability that the humidity at time τ was high Is changed to 0.2. Based on these values, by collecting each information, the probability that the execution condition is satisfied is calculated as follows. In the following, it is assumed that “do nothing” (a2) is selected as the service operation at the current time point.

  When a value is obtained from a certain information acquisition device, the service execution condition satisfaction evaluation value is the service execution condition using the posterior probability when that value is obtained for each value that can be obtained from the information acquisition device. It is obtained by checking a value that satisfies the condition (in the above example, the determination result changes from a2 to a1) and taking the sum of the probabilities of taking that value.

  First, a specific value is set as evidence for the information acquisition device that calculates the service execution condition satisfaction evaluation value. For example, set the value H = True. Subsequently, a service execution condition determination result is obtained when the humidity sensor information and the weather forecast information are obtained. This condition determination is performed by selecting an action with the highest expected utility using a probability value (obtained by an inference algorithm) when evidence is obtained.

Here, instead of the expected utility calculated using the normal equation (1), the expected utility is calculated using the following equation.

here,

Is the probability of the value ri, obtained using a prediction and inference algorithm, and is the prior probability of ri when no prediction is used. In this way, the reciprocal ratio of the prior probabilities of normal probability values is taken for the probabilities obtained using the prediction and inference algorithms. Thereby, the magnitude of the probability that the value at the time τ of interest is likely to be ri again can be evaluated.

  For example, the service execution condition determination when the humidity information is obtained in the service execution condition (1) is executed as follows.

When humidity is high (H = True):
EU (a1 | H = True) = (0.6035 / 0.2134) × 10 + (0.3965 / 0.7866) × 6 = 4.3119
EU (a2 | H = True) = (0.6035 / 0.2134) × (-7) + (0.965 / 0.7866) × 9 = -8.755
In this case, it is recommended to bring an umbrella as the result of the service execution condition determination (a1). Here, “0.2134” indicates the prior probability that it will rain (P (R = True)). Further, “0.7866” is obtained as a prior probability (P (R = False)) that it will not rain, that is, 1-0.2134. “0.6035” here is based on the assumption that H (high humidity) is True, H (high humidity) is True and W (weather forecast is rain) is True, and H is True and W is False. For each, the sum of the probabilities that R (raining) is true is obtained. That is, it is obtained as 0.01 × 0.95 + 0.99 × 0.6 = 0.6035. On the other hand, “0.3965” is obtained as 1−0.6035.

When humidity is low (H = False):
EU (a1 | H = False) = (0.0288 / 0.2134) × 10 + (0.9712 / 0.7866) × 6 = 8.758
EU (a2 | H = False) = (0.0288 / 0.2134) × (-7) + (0.9712 / 0.7866) × 9 = 10.167
In this case, it is not recommended to carry an umbrella (a2). “0.0288” here is based on the premise that H (high humidity) is False, H (high humidity) is False and W (weather forecast is rain) is True, and H is False and W is False. The sum of the probabilities that R (raining) is true for each pair. That is, it is obtained as 0.01 × 0.9 + 0.99 × 0.02 = 0.0288. On the other hand, “0.9712” is obtained as 1−0.0288.

  Since the prediction result shows that the probability of high humidity is 0.1 and the low probability is 0.9, the decision is made by obtaining evidence about whether the humidity is high (ie, collecting information from the humidity sensor) In addition, the probability that the service operation changes to a1 (confirms that the service execution condition is satisfied) is obtained as the probability that the humidity increases.

  Therefore, the service execution condition satisfaction evaluation value obtained by collecting information on humidity at time τ is 0.1.

  Similarly, the probability that the service execution condition is satisfied is obtained by checking the weather forecast.

If the weather forecast is rain (W = True):
EU (a1 | W = True) = (0.905 / 0.2134) × 10 + (0.095 / 0.7866) × 6 = 43.135
EU (a2 | W = True) = (0.905 / 0.2134) × (-7) + (0.095 / 0.7866) × 9 = -29.017
In this case, it is recommended to bring an umbrella as the result of the service execution condition determination (a1). “0.905” here is a premise that W (weather forecast is rain) is True, H (high humidity) is True and W (weather forecast is rain) is True, and H is False and W is It is obtained by the sum of the probabilities that R (raining) will be True for each True pair. That is, it is obtained as 0.1 × 0.95 + 0.9 × 0.9 = 0.905. On the other hand, “0.095” is obtained as 1−0.905.

If the weather forecast is not rain (W = False):
EU (a1 | W = False) = (0.24 / 0.2134) × 10 + (0.76 / 0.7866) × 6 = 17.047
EU (a2 | W = False) = (0.24 / 0.2134) × (-7) + (0.76 / 0.7866) × 9 = 0.823
In this case, it is recommended that the service execution condition determination result also has an umbrella (a1). “0.24” here is based on the premise that W (weather forecast is rain) is False, H (high humidity) is True and W (weather forecast is rain) is False, and H is False and W is It is obtained by the sum of the probabilities that R (raining) is true for each of the false pairs. That is, it is obtained as 0.1 × 0.6 + 0.9 × 0.02 = 0.24. On the other hand, “0.76” is obtained as 1−0.24.

  Since the prediction result shows that the probability that the weather forecast is rain is 0.01, and the probability that it is not rain is 0.99, the service operation is performed when making a decision by obtaining evidence about whether the weather forecast is rain or not. The probability of changing to a1 (confirming that the service execution condition is satisfied) is obtained as the sum of the probability that the weather forecast will be rain and the probability that it is not rain.

  Therefore, the service execution condition satisfaction evaluation value obtained by collecting weather forecast information at time τ is 1.0.

  Similarly, the service execution condition satisfaction evaluation value of each information acquisition device in the service execution conditions (2) and (3) shown in FIGS. 3 and 4 can be derived. The service execution condition (2) refers to an indoor positioning device that obtains information about whether or not the user is in a conference room and an in-house system that obtains information about whether or not there is a conference reservation for the user to participate. Also in this example, it is assumed that a conditional probability table is given as in the case of the service execution condition (1). By referring to the value, the probability that the service execution condition is satisfied when information is obtained from each information acquisition device can be calculated.

  Next, the threshold comparison unit 13 and the information collection control unit 14 will be described.

  The threshold value comparison unit 13 inputs a probability that the service execution condition is satisfied when information is collected from each information acquisition device at each time point after the current time obtained from the execution condition satisfaction evaluation value deriving unit 12. Compare them with a defined threshold. Subsequently, the information acquisition device whose satisfaction probability is higher than the threshold at the nearest time after the current time and the time are transmitted to the information collection control unit 14.

  The information collection control unit 14 sends an information notification instruction to the input information collection control device at the time input from the threshold comparison unit 13.

  As an example, the operation in service example (1) is shown. The threshold set in the threshold comparison unit 13 is set to 0.5. The execution condition satisfaction evaluation value deriving unit 12 obtains an execution condition satisfaction evaluation value of each information acquisition device at a future time after the current time, and inputs it to the threshold comparison unit. The probability that the service execution condition for obtaining information from each of the humidity sensor and the weather forecast site before time τ is 0.5 or less, the condition satisfaction evaluation value of the humidity sensor at time τ is 0.1, and the condition of the weather forecast site is satisfied When the evaluation value is 1.0, since an information acquisition device having a condition satisfaction evaluation value exceeding a preset threshold value 0.5 appears for the first time at time τ, information is collected from a humidity sensor having a value higher than this value at time τ. To do.

(Service provision method)
Next, a service providing method according to the present embodiment will be described with reference to FIGS.

  First, in FIG. 6, the information collection control device 10 acquires service execution conditions and measurement values (S101), and sets the time (τ) for calculating the service execution condition satisfaction evaluation value as the current time (S102). In this embodiment, τ is incremented by unit time from the current time, and the service execution condition satisfaction evaluation value is calculated using the predicted value at each time, so that there is a probability that the service execution condition is satisfied by collection from the information collection device. First, a time that is equal to or greater than the threshold is obtained.

  Next, the prediction unit 11 predicts information at time τ based on information collected in the past from the information acquisition device 50 (S103). Here, a case will be described as an example in which position information is acquired using position positioning means such as GPS provided in a portable terminal carried by the user. It is assumed that the user exists at a coordinate [20m, 30m] in a predetermined two-dimensional coordinate system at the previous positioning and exists at a coordinate [20m, 40m] at the previous positioning. At this time, the moving speed can be calculated from the coordinates at the time of the previous and previous measurement, and the current position can be predicted from the previous position. With this prediction, the probability distribution of the user's position is obtained such that the probability of being within a circle of radius 20 m of [20 m, 50 m] at time τ is 0.8, and the probability of being other than that is 0.2.

  Next, using the service execution condition 21 stored in the service execution condition storage unit 20 and the value predicted by the prediction unit 11, the execution condition satisfaction evaluation value deriving unit 12 receives information from each information acquisition device in each service. An execution condition satisfaction evaluation value is obtained when S is obtained (S104). A flowchart of the execution condition satisfaction evaluation value derivation process is shown in FIG. This derivation process will be described later.

  Next, the threshold value comparison unit 13 compares the derived execution condition satisfaction evaluation value of each information acquisition device with a predetermined threshold value (S105). If there is an information acquisition device having an execution condition satisfaction evaluation value equal to or greater than the threshold, the information collection control unit 14 determines to acquire information from the information acquisition device at time τ, and ends the procedure (S106 and S106). S107). If there is no information acquisition device whose execution condition satisfaction evaluation value is equal to or greater than the threshold, whether or not there is an information acquisition device whose execution condition satisfaction evaluation value at the future time is equal to or greater than the threshold by advancing time τ The determination is continued (S106 and S108).

  Next, the execution condition satisfaction evaluation value derivation process in FIG. 7 will be described.

  First, the information collection control device 10 acquires service execution conditions and measurement values (S201). Then, a set of all information acquisition devices used by the service execution condition is set as Y (S202), and an information acquisition device for which the satisfaction probability of the current execution condition is calculated is set as f (S203).

  Subsequently, for each node corresponding to the Y element other than f, the prior probability of the node in the Bayes decision tree is updated based on the prediction result at time τ (S204). Subsequently, a set of state values of the node corresponding to f is set as V (S205), and the probability that the determination result of the service execution condition changes according to the following procedure is obtained (S206). A flowchart of this probability acquisition process is shown in FIG. This acquisition process will be described later.

  Then, the information collection control device 10 sets the obtained value as an execution condition satisfaction evaluation value (S207).

  Next, the probability acquisition process in FIG. 8 will be described.

  First, one value is extracted from V and set to v (S301). Next, evidence that the value is v is set in the node corresponding to f (P (f = v) = 1.0) (S302).

  Then, Expression (2) is used for the expected utility calculation, and the determination result of the service execution condition is obtained based on these values. This condition determination is performed by selecting an operation with the highest expected utility using a probability value (obtained by an inference algorithm) when evidence is obtained. Here, a predicted value at time τ is used for a node corresponding to an information acquisition device other than V.

  Subsequently, when this determination result is different from the current determination result, it is understood that the service execution condition is satisfied when the value is obtained. Therefore, the probability of taking the value is obtained from the prediction result of the prediction unit 11 and added to the result R. (S304 and S305).

  In this way, for all the state values (S306), it is determined whether or not the execution condition determination result when the value is obtained is changed, and the sum of the probabilities of taking the changing value is obtained. The sum R of the probabilities becomes the probability that the service execution condition is satisfied (the determination result changes) by collecting information from V.

(Function and effect)
According to the information collection control device and the information collection control method according to the present embodiment, the service execution condition satisfaction evaluation value in the case of collecting information from the information acquisition device used in the service execution condition is derived, and the derived probability is calculated. By controlling the information collection from the information acquisition device based on the above, it becomes possible to control the information collection according to the probability that the service execution condition is satisfied, thereby reducing the number of information collection times and the number of information acquisition times of the information acquisition device. Can be planned. Thereby, it is possible to increase the number of services that the system can provide simultaneously while maintaining the total cost of the entire system.

  The service execution condition is described in the form of a Bayes decision tree, and the execution condition satisfaction evaluation value deriving unit updates the prior probabilities of the nodes in the Bayes decision tree based on the prediction result by the prediction unit, and derives an evaluation value. When the candidate corresponding to the information collection control device uses each candidate value that can be collected from the information collection control device, evidence is set for the node, and the candidate value is set as evidence in the previous candidate value, the update is performed. Based on the probability distribution of the obtained nodes, a value having a high probability that the service execution condition is satisfied is checked, and the sum of the probabilities collected from the information collection control device from which the value derives the evaluation value is used as the evaluation value.

  Therefore, when analyzing the service execution condition described in the form of the Bayes decision tree, the execution condition satisfaction evaluation value can be immediately obtained by analyzing the Bayes decision tree.

(Modification)
In the above embodiment, the information collection control device and the service execution condition storage unit are arranged in a single server. However, as shown in FIG. 9, the information collection control device 10 and the service execution condition storage are stored. A configuration in which the unit 20 and the information acquisition device 50 are arranged in a single terminal device 200 may be adopted. In this case, examples of the information acquisition device 50 include a GPS device provided in the terminal device 200, indoor positioning means, and a temperature sensor.

(Other embodiments)
Although the present invention has been described according to the above-described embodiments, it should not be understood that the descriptions and drawings constituting a part of this disclosure limit the present invention. From this disclosure, various alternative embodiments, examples and operational techniques will be apparent to those skilled in the art.

  For example, in the above-described embodiment, the information collection control device 10 and the service execution condition storage unit 20 have been described as being arranged in the same server or terminal device. However, the information collection control device 10 and the service execution condition storage unit 20 May be arranged in different apparatuses. That is, the information acquisition device 50, the information collection control device 10, and the service execution condition storage unit 20 may be arranged separately.

  As described above, the present invention naturally includes various embodiments not described herein. Therefore, the technical scope of the present invention is defined only by the invention specifying matters according to the scope of claims reasonable from the above description.

It is a block diagram of the service providing system according to the present embodiment (No. 1). It is a block diagram of the service providing system according to the present embodiment (No. 2). It is a figure which shows the service execution conditions of the preparation recommendation service of the umbrella which concerns on this embodiment. It is a figure which shows the service execution conditions of the automatic manner mode service which concerns on this embodiment. It is a figure which shows the service execution conditions of the service completion notification service which concerns on this embodiment. It is a flowchart which shows the service provision method which concerns on this embodiment (the 1). It is a flowchart which shows the service provision method which concerns on this embodiment (the 2). It is a flowchart which shows the service provision method which concerns on this embodiment (the 3). It is a block diagram of a service providing system according to a modification.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Information collection control apparatus 11 ... Prediction part 12 ... Execution condition satisfaction evaluation value derivation part 13 ... Threshold comparison part 14 ... Information collection control part 20 ... Service execution condition storage part 21 ... Service execution condition 30 ... Communication network 50 ... Information acquisition Device 100 ... Server 101 ... CPU
DESCRIPTION OF SYMBOLS 102 ... Input part 103 ... Output part 104 ... Data holding part 105 ... Program holding part 200 ... Terminal device

Claims (4)

An information collection control device arranged in a service providing system that provides a service at a time when the service execution condition is satisfied based on information collected from one or more information acquisition devices and one or more service execution conditions ,
Prediction means for predicting information collected from the information acquisition device at a future time based on information collected from the information acquisition device in the past;
The service execution condition is satisfied when information is collected from one or more information acquisition devices used in the service execution condition based on the analysis result of the service execution condition using the prediction result by the prediction unit. Execution condition satisfaction evaluation value deriving means for deriving an evaluation value based on the probability of being
Threshold value comparing means for comparing the evaluation value with a predetermined threshold value, and identifying an information acquisition device having the evaluation value higher than the threshold value;
Information collection control means for transmitting an information collection instruction at the time to the information acquisition apparatus having the evaluation value higher than the threshold at the nearest future time among the information acquisition apparatuses specified by the threshold comparison means. An information collection control device comprising: The service execution condition is described in the form of a Bayesian decision tree,
The execution condition satisfaction evaluation value deriving unit updates the prior probability of the node in the Bayesian decision tree based on the prediction result by the prediction unit, and collects the information in a node corresponding to the information collection control device that derives the evaluation value The evidence is set for the node using each candidate value that can be collected from the control device, and the service is based on the updated probability distribution of the node when the candidate value is set as evidence among the previous candidate values. 2. The information according to claim 1, wherein a value having a high probability that the execution condition is satisfied is checked, and the sum of the probabilities collected from the information collection control device that derives the evaluation value is used as the evaluation value. Collection controller. An information collection control method arranged in a service providing system that provides a service at a time when the service execution condition is satisfied based on information collected from one or more information acquisition devices and one or more service execution conditions. ,
Predicting information collected from the information acquisition device at a future time based on information collected from the information acquisition device in the past;
When information is collected from one or more information acquisition devices used in the service execution condition based on the analysis result of the service execution condition using the prediction result in the prediction step, the service execution condition is Deriving an evaluation value based on a probability of being satisfied;
Performing a comparison between the evaluation value and a predetermined threshold value, and identifying an information acquisition device in which the evaluation value is higher than the threshold value;
Transmitting an information collection instruction at the time to the information acquisition device whose evaluation value is higher than the threshold value at the nearest future time among the specified information acquisition devices. Collection control method. The service execution condition is described in the form of a Bayesian decision tree,
In the deriving step, the prior probability of the node in the Bayesian decision tree is updated based on the prediction result by the prediction unit, and collected from the information collection control device to a node corresponding to the information collection control device from which the evaluation value is derived If the candidate value is set as evidence for each node and the candidate value is set as evidence among the previous candidate values, the service execution condition is satisfied based on the updated probability distribution of the node. 4. The information collection control method according to claim 3, wherein a value having a high probability of being checked is examined, and a sum of probabilities collected from the information collection control device that derives the evaluation value is used as the evaluation value.

Images