WO2016125273A1 - Information processing device, information processing method, and information processing program - Google Patents

Abstract

An information storage unit (101) stores event information in which are delineated a cause event occurrence time when a cause event has occurred to an event occurrence subject, a result event occurrence time when a result event has occurred to the event occurrence subject, and an environment impact time when an environment with which the event occurrence subject is associated has begun having an impact on the event occurrence subject. A computation formula generating unit (102) generates, using the result event occurrence time and the environment impact time, an environment impact computation formula for computing a value of an environment impact coefficient which is a saturation coefficient of the impact from the environment, and, using the cause event occurrence time and the result event occurrence time, generates a cause-impact computation formula for computing a value of a cause-impact coefficient which is an attenuation coefficient of the impact from the cause event. A coefficient value computation unit (103) sets a restriction condition and a target function to the environment impact computation formula and the cause-impact computation formula, and computes the value of the environment impact coefficient and the cause-impact coefficient from the environment impact computation formula and the cause-impact coefficient formula to which the restriction condition and the target function are set.

Description

Information processing apparatus, information processing method, and information processing program

The present invention relates to a technique for analyzing a causal relationship between events.

When a causal event and a result event are given as events that are assumed to have a causal relationship, evaluating the influence of the cause event on the occurrence of the result event and predicting the occurrence of the result event is an analysis of corporate activities, etc. Has an important meaning.
For example, when considering a device modulation as a cause event and a device failure as a result event, if a device failure can be estimated from a prior modulation, preventive measures can be taken to prevent losses due to sudden failures. it can.
In addition, when an advertising strategy is considered with direct mail transmission to a customer as a cause event and product purchase as a result event, it is required to send direct mail to an appropriate customer and lead to product purchase with high probability.
For this reason, various approaches are used to quantitatively evaluate the influence of the cause event on the occurrence of the result event.

Patent Document 1 discloses a method for calculating the influence of a cause event and a hidden event derived from the cause event and predicting the occurrence of a result event.
In Patent Document 2, an event that has occurred is accumulated, time-series data including temporal changes and a sequence that is order information are extracted from the accumulated event, taking into account temporal importance, A method for calculating the degree of similarity is disclosed.

Japanese Patent No. 5413240 JP 2002-207755 A

In the methods of Patent Document 1 and Patent Document 2, the occurrence of a result event is predicted in consideration of the temporal variation of the influence from the cause event.
However, in actuality, it is assumed that the influence that is constantly exerted from the environment contributes to the occurrence of the result event in addition to the cause event.
For example, in maintenance service operations, the education of maintenance personnel is regarded as a cause event, and the improvement of the maintenance staff's technical capabilities is regarded as a result event, and the impact of education and education on the improvement of technical capabilities can be evaluated. Done.
However, it should be taken into consideration that the technical skills of maintenance work will gradually improve not only through temporary training but also through daily maintenance work activities.
The degree of improvement varies greatly depending on the environment such as the maintenance target.

For each method of evaluating the occurrence probability of a result event by simultaneously considering two types of effects that change (decay) over time from multiple cause events that occur temporarily and effects that are received from an environment that always exists, Since there are a variety of influence models from cause events, it is difficult to properly model all the cause events, so an efficient method has not been proposed.

The present invention has been made in view of such circumstances, and has as its main object to quantify the influence of a cause event that decays with time and the influence of an environment that always exists.

An information processing apparatus according to the present invention includes:
The cause event occurrence time that is the time when the cause event occurred in the event occurrence target, the result event occurrence time that is the time when the result event occurred in the event occurrence target, and the environment to which the event occurrence target belongs is the event occurrence target. An information storage unit for storing event information in which an environmental impact time that is a time at which an influence starts is described;
An environmental impact calculation formula that is a calculation formula for calculating a value of an environmental impact coefficient that is a saturation coefficient of an environmental impact that is an impact from the environment is generated using the result event occurrence time and the environmental impact time. , Using the cause event occurrence time and the result event occurrence time as a cause effect calculation formula that is a calculation formula for calculating a value of a cause influence coefficient that is an attenuation coefficient of a cause influence that is an influence from the cause event A calculation formula generation unit for generating
A constraint condition and an objective function are set in the environmental impact calculation formula and the cause impact calculation formula, and the environmental impact calculation formula and the cause impact calculation formula in which the constraint condition and the objective function are set, A coefficient value calculating unit for calculating the value of the environmental influence coefficient and the value of the cause influence coefficient;

According to the present invention, the value of the cause influence coefficient that is the attenuation coefficient of the cause influence that is the influence from the cause event and the value of the environment influence coefficient that is the saturation coefficient of the environment influence that is the influence from the environment are calculated. It is possible to quantify the influence of the cause event that decays with time and the influence from the environment that always exists.

FIG. 3 is a diagram illustrating a functional module configuration example of the information processing apparatus according to the first embodiment. FIG. 3 is a flowchart showing an operation example of the information processing apparatus according to the first embodiment. FIG. 6 is a diagram illustrating an example of cause event information according to the first embodiment. FIG. 6 is a diagram showing an example of result event information according to the first embodiment. FIG. 4 is a diagram illustrating an example of environment information according to the first embodiment. FIG. 6 is a diagram illustrating an example of cause influence coefficient information according to the first embodiment. FIG. 4 is a diagram showing an example of environmental impact coefficient information according to the first embodiment. FIG. 6 is a diagram showing an example of new cause event information according to the first embodiment. FIG. 6 is a diagram showing an example of result event candidate information according to the first embodiment. The figure which shows the relationship of the cause event which concerns on Embodiment 1, a result event, and an environment. FIG. 6 is a diagram illustrating a calculation example of a result occurrence index according to the first embodiment. The figure which shows the example of the value at the time of the result event generation | occurrence | production of the result generation | occurrence | production index which concerns on Embodiment 1. FIG. FIG. 6 is a diagram illustrating an example of program code corresponding to processing of a result event candidate calculation unit according to the first embodiment. FIG. 3 is a diagram illustrating a hardware configuration example of the information processing apparatus according to the first embodiment.

Embodiment 1 FIG.
In the present embodiment, the degree of occurrence of a result event is estimated by simultaneously considering, on the same scale, two types of effects, which are attenuating over time from a cause event that occurs temporarily and an effect received from an environment that always exists. The structure to perform is demonstrated.

*** Explanation of configuration ***
FIG. 1 shows a functional module configuration example of the information processing apparatus 100 according to the present embodiment.
Before describing the details of the components of the information processing apparatus 100, the operation principle of the information processing apparatus 100 will be described.

The information processing apparatus 100 formulates event correlation estimation, thereby estimating the correlation between the cause event and the influence on the result event from the environment.
The information processing apparatus 100 according to the present embodiment manages the cause event {Ci}, the result event {Ri}, and the event occurrence target (thing, person) {Oi} in association with each other.
The cause event is, for example, modulation of the device, the result event is failure of the device, and the event generation target is the device.
In the present embodiment, it is assumed that the cause event {Ci} and the result event {Ri} exist on the time axis for each event generation target (thing, person) {Oi}.
Both the cause event and the result event can be classified into several types, and the event occurrence target is assumed to belong to several types of environments {Ei} (FIG. 10).

The event generation target Oi, it is assumed that the time t _i as a reference exists.
Environmental impacts (hereinafter, environmental impacts are also referred to as “environmental impacts”) accumulate from the reference time and become saturated at the same time.
That is, the environment begins to affect the event occurrence target Oi from time t _i (hereinafter, time t _i is also referred to as environment influence time).
It is assumed that the environmental impact is given by Aj (1-exp (−Bj (t _qi −t _i )).
The time t _qi is the time when the result event occurs (hereinafter, the time t _qi is also referred to as the result event occurrence time).
The variable Aj is a variable indicating the maximum amount of environmental influence (or the amount of influence at the time when a certain elapsed time has passed without being limited to the maximum amount) as the intensity of the saturated environmental influence.
As described above, the variable Aj is a coefficient of a reference value for the strength of environmental impact, and is also referred to as an environmental impact reference coefficient.
The variable Bj is a variable indicating the increasing speed of the saturation effect.
Thus, the variable Bj is a coefficient of speed at which the strength of the environmental influence is saturated, and is also referred to as an environmental influence saturation speed coefficient.
The variables Aj and Bj are both environmental impact saturation coefficients, and are collectively referred to as environmental impact coefficients.
The above Aj (1-exp (−Bj (t _qi −t _i )) is a calculation formula (saturation function) for calculating the value of the environmental influence coefficient (variable Aj, variable Bj). This is called a calculation formula.
Note that the variable Aj and the variable Bj are determined for each combination of environment type and result event type.

It is assumed that the influence from the cause event (hereinafter, the influence from the cause event is also referred to as “cause influence”) attenuates with time.
For example, when the cause event occurrence time at which the cause event occurs is t _ki , the cause influence is assumed to decrease at a ratio of Fj exp (−Gj (t _qi −t _ki )).
The variable Fj is a variable that indicates the maximum amount of the cause effect (or the amount of the cause effect at a certain point of time without being limited to the maximum amount) as the strength of the cause effect before the cause effect is attenuated. .
Thus, the variable Fj is a coefficient of the reference value for the strength of the cause influence, and is also referred to as a cause influence reference coefficient.
The variable Gj is a variable indicating the rate of decrease when the cause influence is attenuated.
As described above, the variable Gj is a coefficient of speed at which the strength of the cause influence is attenuated, and is also referred to as a cause influence attenuation speed coefficient.
The variables Fj and Gj are both cause-effect attenuation coefficients, and are collectively referred to as cause-effect coefficients.
The above Fj exp (−Gj (t _qi -t _ki )) is a calculation formula (convergence function) for calculating the value of the cause influence coefficient (variable Fj, variable Gj), and is called the cause influence calculation formula. .
The variable Fj and the variable Gj are determined for each combination of the cause event type, the result event type, and the type of environment to which the event occurs.

As an example of modeling the decay of the causal influence by an exponential function, for example, there is a curve called “Ebinghouse forgetting curve” related to forgetting of the learned content.
This curve indicating the degree to which forgetting progresses after learning (the influence of learning attenuates) is also approximated by an exponential function.
References: Niki et al. “Learning neural model expressing long-term memory at synapses: Representation and analysis of memory / learning / self-organization by H / neurons”, IEICE Transactions A, Vol. J64-A, no. 11, pp. 948-955

It is assumed that a result occurrence index Pi of an event occurrence target is determined by accumulation of past cause events received by the event occurrence target and influences from the environment.
The result occurrence index Pi is the occurrence probability of the result event.
It is assumed that the result occurrence index Pi exceeds the separately set result occurrence index threshold value h at the time when the result event occurs.
When attention is paid to the formulated value at the time t _qi when the result event occurs, the following inequality is obtained.
h ≦ Pi = Aj (1−exp (−Bj (t _qi −t _i )) + Σ (Fj exp (−Gj (t _qi −t _ki )))

In Equation 1, the values of t _qi , t _i , and t _ki are known, and variables Aj, Bj, Fj, and Gj are unknowns.
Since Expression 1 is obtained for each result event, the information processing apparatus 100 generates simultaneous inequalities using Expression 1 for each result event.
Then, the information processing apparatus 100 obtains the environmental influence coefficients Aj and Bj and the cause influence coefficients Fj and Gj using mathematical programming so that the sum of Pi-h values (threshold difference) is minimized.
As a method for solving mathematical programming, there is a method using a quasi-Newton method.
References: Yamashita “Research and Prospects of Quasi-Newton Method”, Operations Research: Management Science 55 (4), 243-247, 2010.

Then, the information processing apparatus 100 applies the calculated values of the environmental influence coefficients Aj and Bj and the values of the cause influence coefficients Fj and Gj to the above-described equation 1, and determines the result event occurrence candidate for the new cause event. As a result, the occurrence index Pi is calculated.

Next, the functional module configuration of the information processing apparatus 100 according to the present embodiment will be described.
The information processing apparatus 100 includes an information storage unit 101, a calculation formula generation unit 102, a coefficient value calculation unit 103, and a result event candidate calculation unit 104.

The information storage unit 101 is a storage area for storing information.
The information storage unit 101 includes a cause event DB (Database) 1011, a result event DB 1012, an environment information DB 1013, and a new cause event DB 1014.
The cause event DB 1011 stores cause event information illustrated in FIG.
In the cause event information, a cause event occurrence time t _ki that is a time when the cause event has occurred in the event occurrence target is described.
Details of the cause event information will be described later.
The result event DB 1012 stores result event information illustrated in FIG.
In the result event information, a result event occurrence time t _qi that is a time at which a result event has occurred in the event occurrence target is described.
Details of the result event information will be described later.
The environment information DB 1013 stores the environment information illustrated in FIG.
The environment information describes an environment influence time t _i that is a time when the environment to which the event occurrence target belongs starts to affect the event occurrence target.
Details of the environmental information will be described later.
The cause event information, result event information, and environment information are also collectively referred to as event information.
The new cause event DB 1014 stores new cause event information illustrated in FIG.
In the new cause event information, a new cause event that has occurred after the values of the environmental influence coefficients Aj and Bj and the values of the cause influence coefficients Fj and Gj are calculated by a coefficient value calculation unit 103 described later is described.
Details of the new cause event information will be described later.

The calculation formula generation unit 102 is an environmental impact calculation formula: Aj (1-exp (−Bj (t _qi −t _i )), which is a combination of a polynomial function and an exponential function, and a cause influence calculation formula: (Fj exp (−Gj ( t _qi -t _ki )) is generated using the cause event occurrence time t _ki , the result event occurrence time t _qi, and the environmental impact time t _i .
More specifically, the calculation formula generation unit 102 includes an environmental impact formulation unit 1021 and a cause impact formulation unit 1022, and the environmental impact formulation unit 1021 has an environmental impact calculation formula: Aj (1-exp (−Bj (T _qi -t _i )) is generated, and the cause / effect formulation unit 1022 generates a cause / effect calculation formula: (Fj exp (-Gj (t _qi -t _ki )).

The coefficient value calculation unit 103 sets a constraint condition and an objective function in the environmental influence calculation formula and the cause influence calculation formula.
Further, the coefficient value calculation unit 103 calculates the values of the environmental influence coefficients Aj and Bj and the values of the cause influence coefficients Fj and Gj from the environmental influence calculation formula and the cause influence calculation formula in which the constraint condition and the objective function are set. calculate.
The coefficient value calculation unit 103 calculates the values of the environmental influence coefficients Aj and Bj and the values of the cause influence coefficients Fj and Gj that minimize the objective function by mathematical programming.
The constraint condition is h ≦ Pi in Equation 1.
That is, the coefficient value calculation unit 103 sets a constraint condition for the environmental impact calculation formula and the cause impact calculation formula, and generates an inequality of Formula 1.
The objective function is the sum of Pi−h values (threshold difference).
That is, the coefficient value calculation unit 103 minimizes the total sum of Pi-h values (threshold difference) as the objective function.
The coefficient value calculation unit 103 includes a co-occurrence relationship evaluation unit 1031 and a mathematical programming calculation unit 1032. The co-occurrence relationship evaluation unit 1031 generates an inequality of Equation 1, sets an objective function, and a mathematical programming calculation unit. 1032 performs mathematical programming operations.
The values of the environmental influence coefficients Aj and Bj calculated by the coefficient value calculation unit 103 are output to the result event candidate calculation unit 104 as the cause influence coefficient information 300, and the cause influence coefficient Fj calculated by the coefficient value calculation unit 103 is output. And Gj are output to the result event candidate calculation unit 104 as the environmental impact coefficient information 400.

The result event candidate calculation unit 104 determines the environmental effect coefficient when a new cause event occurs after the coefficient value calculation unit 103 calculates the values of the environmental influence coefficients Aj and Bj and the cause influence coefficients Fj and Gj. The values of Aj and Bj are applied to the environmental influence calculation formula, and the values of the cause influence coefficients Fj and Gj are applied to the cause influence calculation formula to identify result event candidates that may occur due to a new cause event.
Further, the result event candidate calculation unit 104 calculates a result occurrence index Pi (occurrence probability) of the result event candidate.
When there are a plurality of result event candidates, the result event candidate calculation unit 104 calculates a result occurrence index Pi for each result event candidate, and ranks the plurality of result event candidates in descending order of the calculated result occurrence index Pi. .
In addition, the result event candidate calculation unit 104 predicts the time when the result event candidate occurs.
Then, the result event candidate calculation unit 104 outputs result event candidate information 500 indicating a list of result event candidates and a time when the result event candidate occurs.

Next, the cause event information of FIG. 3 will be described.
In FIG. 3, the cause ID is an ID (Identifier) of the cause event that has occurred.
The cause ID is expressed as Ci (i = 1, 2, 3,...).
C is an abbreviation for Cause.
The cause type ID is an ID of a cause event type.
The cause type ID is expressed as CTi (i = 1, 2, 3,...).
CT is an abbreviation for Cause Type.
The occurrence date and time is the cause event occurrence time t _ki .
The event generation target ID is an ID of the event generation target in which the cause event has occurred.
The event occurrence target ID is expressed as Oi (i = 1, 2, 3,...).
O is an abbreviation for Object.

Next, the cause event information of FIG. 4 will be described.
The result ID is an ID of a result event that has occurred.
The result ID is expressed as Ri (i = 1, 2, 3,...).
R is an abbreviation for Result.
The result type ID is a result event type ID.
The result type ID is expressed as RTi (i = 1, 2, 3,...).
RT is an abbreviation for Result Type.
The occurrence date and time is the result event occurrence time t _qi .
The event generation target ID is an ID of the event generation target where the result event has occurred.

Next, the environment information in FIG. 5 will be described.
The event generation target ID is an ID of the event generation target where the result event has occurred.
The environment type ID is an ID of the type of environment to which the event generation target in which the result event has occurred belongs.
The environment ID is expressed as ETi (i = 1, 2, 3,...).
ET is an abbreviation for Environment Type.
The reference date and time, is an environmental impact time t _i.
The reference date and time is set for each event occurrence target.

Next, the new cause event information of FIG. 8 will be described.
The cause ID is an ID of a cause event that occurs after the values of the environmental influence coefficients Aj and Bj and the values of the cause influence coefficients Fj and Gj are calculated.
The meanings of the cause type ID, the occurrence date and time, and the event occurrence target ID are the same as those shown in FIG.

Next, an example of the cause influence coefficient information 300 is shown in FIG.
The variable Fj is a cause influence reference coefficient.
The variable Gj is a cause influence attenuation rate coefficient.
The cause type ID is the same as that shown in FIG.
The result type ID is the same as that shown in FIG.
The environment type ID is the same as that shown in FIG.

Next, an example of the environmental impact coefficient information 400 is shown in FIG.
The variable Aj is an environmental impact standard coefficient.
The variable Bj is an environmental influence saturation speed coefficient.
The environment type ID is the same as that shown in FIG.
The result type ID is the same as that shown in FIG.

Next, an example of the result event candidate information 500 is shown in FIG.
The result candidate number is a number for identifying a result event candidate.
The result occurrence index is an index obtained by numerically converting the likelihood of occurrence of a result event.
As the result occurrence index is larger, a result event is more likely to occur.
The expected result occurrence period start date and time is the date and time when the expected result occurrence period starts.
The expected result occurrence period end date and time is the date and time when the expected result occurrence period ends.
The expected result occurrence period is a period in which the result event candidate calculation unit 104 assumes that the value of the result occurrence index Pi exceeds the result occurrence index threshold value h and that a result event may occur.

*** Explanation of operation ***
The information processing apparatus 100 according to the present embodiment formulates the influence that decays with time according to the cause influence coefficient, based on the occurrence of the cause event, and the influence that is always affected according to the type of environment by the environment influence coefficient. Formulate to settle.
From the state in which the cause influence coefficient and the environmental influence coefficient are indefinite, the information processing apparatus 100 causes the cause influence coefficient and the cause influence coefficient to increase so that the ratio at which the occurrence probability of the result event exceeds the threshold when the result event occurs is maximized. Calculate the environmental impact factor.
The cause influence coefficient indicates the strength of the relationship between the cause event and the result event, and the environmental impact coefficient indicates the strength of the relationship between the environment and the result event.

FIG. 2 shows an operation example of the information processing apparatus 100 according to the present embodiment.
Hereinafter, an operation example of the information processing apparatus 100 according to the present embodiment will be described with reference to FIG.
The procedure shown in FIG. 2 corresponds to an example of the information processing method and information processing program of the present application.

In S101, the environmental impact formulation unit 1021 reads each record of the result event information from the result event DB 1012 and records the environment information including the same event occurrence target ID as the event occurrence target ID described in each read record. Is read from the environment information DB 1013 (information read processing).

In S102, the environmental impact formulation unit 1021 _{obtains the} result event occurrence time t _qi (occurrence date and time of FIG. 4) from the record of the result event information and the record of the environment information with the same event occurrence target ID for each event occurrence target ID. And the environmental impact time t _i (reference date and time in FIG. 5), and using the event occurrence time t _qi and the environmental impact time t _i as a result of the extraction, an environmental impact calculation formula: Aj (1-exp (−Bj ( t _qi -t _i )) is generated (calculation formula generation process).
Then, the environmental impact formulation unit 1021 outputs the generated environmental impact calculation formula and the record of the result event information used for generating the environmental impact calculation formula to the cause impact formulation unit 1022.

Next, the cause / effect formulation unit 1022 reads from the cause event DB 1011 a record of cause event information in which the event occurrence target ID used to generate the environmental impact calculation formula by the environmental impact formulation unit 1021 is described (information Read processing).
Further, in S103, the cause influence formulation unit 1022 determines the cause event occurrence time t _ki (occurrence date and time in FIG. 3) and the result event occurrence from the cause event information record and the result event information record having the same event occurrence target ID. Time t _qi (occurrence date and time in FIG. 4) is extracted, and using the extracted cause event occurrence time t _ki and result event occurrence time t _qi , a cause influence calculation formula: (Fj exp (−Gj (t _qi −t _ki )) is generated (calculation formula generation process).
Then, the cause / effect formulation unit 1022 outputs the environmental impact calculation formula and the cause / effect calculation formula to the co-occurrence relationship evaluation unit 1031.

Next, in S104, the co-occurrence relationship evaluation unit 1031 calculates the values of the environmental influence coefficients Aj and Bj and the values of the cause influence coefficients Fj and Gj (coefficient value calculation processing).
More specifically, the co-occurrence relationship evaluation unit 1031 calculates the environmental impact calculation formula: Aj (1-exp (−Bj (t _qi −t _i )) and the cause impact calculation formula: (Fj) for the common event occurrence target ID. exp (−Gj (t _qi -t _ki )) is combined to generate the inequality of the above equation 1, and this inequality is used as a constraint, and the result occurrence index Pi and the result occurrence index threshold h are used for the convergence of the coefficients. As a result, a solution is obtained with the objective function of minimizing the sum of each event (result ID) as a result of the difference.
The result occurrence index threshold value h is input as the result occurrence index threshold value 200.
In order to obtain a solution, the mathematical programming calculation unit 1032 performs arithmetic processing by mathematical programming.
The values of the environmental influence coefficients Aj and Bj and the values of the cause influence coefficients Fj and Gj are obtained by the arithmetic processing by the mathematical programming method of the mathematical programming calculation unit 1032.

Next, when a new cause event is added to the result event candidate calculation unit 104 in S105, the result event candidate calculation unit 104 outputs a result event candidate that may occur due to the new cause event.
More specifically, the result event candidate calculation unit 104 converts the values of the environmental impact coefficients Aj and Bj obtained in S103 into an environmental impact calculation formula: Aj (1-exp (−Bj (t _qi −t _i )). Substituting the values of the cause influence coefficients Fj and Gj into the cause influence calculation formula: (Fj exp (−Gj (t _qi −t _ki )).
As a result the event candidate calculation unit 104, environmental environmental impact time _{t i} the environmental impact calculation formula of the event generation target a new cause event has occurred _{belongs: Aj (1-exp (-Bj} (t qi -t i) ) And the cause event occurrence time t _ki of the new cause event is substituted into the cause influence calculation formula: (Fj exp (−Gj (t _qi −t _ki )).
A result occurrence index Pi for an event occurrence target ID in which a new cause event has occurred is calculated for each result type ID and for each time according to Equation 2 below.
Pi = Aj (1−exp (−Bj (t _qi −t _i )) + Σ (Fj exp (−Gj (t _qi −t _ki ))) Equation 2
Here, “calculate by result type ID and time” will be described.
“By result type ID” means that the result event candidate calculation unit 104 calculates the result occurrence index Pi for all RTi of the result type ID {RT1, RT2,.
For “calculate by time”, a candidate calculation maximum date and time t_max (for example, one year later) and a candidate calculation date and time interval t_interval (for example, one month) are given as inputs.
In the calculation by time, the result event candidate calculation unit 104 calculates the value of the result occurrence index Pi at each time point (t = t_now + s × t_interval) when a constant multiple of t_interval has elapsed from the current date and time t_now.
Here, s is a natural number ranging from 1 to (s_max), and (s_max) is a maximum natural number not exceeding {(t_max−t_now) ÷ t_interval}.
“Calculate by result type ID and time” corresponds to performing the process of FIG. 13 when the intermediate calculation result is stored in temp_Pi [et, s].

When the value of the calculated result occurrence index Pi exceeds the result occurrence index threshold value h and takes a maximum value, the result event candidate calculation unit 104 sets the calculated result occurrence index Pi to the corresponding result type ID. As a result, it is considered that the event occurs.
The result event candidate calculation unit 104 calculates the value of the result occurrence index Pi for all result type IDs, and outputs a list in which the result type IDs are arranged in descending order of the result occurrence index Pi as result event candidate information (FIG. 9). To do.

In addition, the calculation formula generation unit 102 generates a new result event after the coefficient value calculation unit 103 calculates the value of the environmental influence coefficient (Aj, Bj) and the value of the cause influence coefficient (Fj, Gj). In addition, the environmental impact calculation formula and the cause impact calculation formula may be updated by reflecting the time of occurrence of a new result event.
In this case, the coefficient value calculation unit 103 sets the above-described constraint condition and objective function in the updated environmental effect calculation formula and the updated cause effect calculation formula by the calculation formula generation unit 102, and the constraint condition and the objective function are set. A new environmental impact coefficient (Aj, Bj) value and cause influence coefficient (Fj, Gj) value are calculated from the updated environmental impact calculation formula and the updated causal impact calculation formula in which the function is set. To do.

FIG. 10 is a diagram schematically illustrating a cause event and a result event that occur with the passage of years for each event generation target.
FIG. 10 shows that the event generation target with event generation target ID: O1 (hereinafter referred to as event generation target O1) belongs to the environment with environment type ID: ET1 (hereinafter referred to as environment ET1).
In addition, a cause event with cause ID: C1 (hereinafter referred to as cause C1) occurs at time t _k1 in the event occurrence target O1, and a cause event with cause ID: C2 (hereinafter referred to as cause C2) occurs at time t _k2. It shows that you are doing.
Further, it is indicated that a result event of result ID: R1 (hereinafter referred to as result R1) has occurred in event generation target O1 at time _tq1 .
The cause type ID of the cause C1 is CT1, the cause type ID of the cause C2 is CT2, and the result type ID of the result R1 is RT1.
FIG. 10 shows that the event generation target of event generation target ID: O2 (hereinafter referred to as event generation target O2) belongs to the environment ET1.
In addition, a cause event of cause ID: C3 (hereinafter referred to as cause C3) occurs in event generation target O2 at time t _k3, and a result event of result ID: R2 (hereinafter referred to as result R2) occurs at time t _q2. It shows that you are doing.
The cause type ID of the cause C3 is CT1, and the result type ID of the result R2 is RT1.
FIG. 10 shows that the event generation target of event generation target ID: O3 (hereinafter referred to as event generation target O3) belongs to the environment of environment type ID: ET2.
Further, it is indicated that a cause event of cause ID: C4 (hereinafter referred to as cause C4) is occurring in the event occurrence target O3 at time _tk4 .
The cause type ID of the cause C4 is CT2.

FIG. 11 is a diagram schematically illustrating calculating the result occurrence index P1 from the following two types of influences ((1) and (2)) on the event generation target O1.
(1) Influence from the cause C1 and the cause C2 (2) Influence from the environment ET1 to which the event occurrence target O1 belongs A graph 1101 in FIG. 11 schematically represents an attenuation curve of the influence from the cause C1.
A graph 1102 in FIG. 11 schematically represents an attenuation curve of the influence from the cause C2.
A graph 1103 in FIG. 11 schematically represents a saturation curve of the influence from the environment ET1.
A graph 1104 in FIG. 11 schematically represents a time transition of the result occurrence index P1, and is obtained by superimposing the graph 1101, the graph 1102, and the graph 1103.
That is, the result occurrence index P1 is an aggregation result of the influence from the cause C1, the influence from the cause C2, and the influence from the environment ET1.
The value of the result occurrence index P1 is given by the following equation.
P1 (t) = A ₁ (1-exp (−B ₁ (t−t ₁ ))) + F ₁ exp (−G ₁ (t−t _k1 )) + F ₂ exp (−G ₂ (t−t _k2 ) )
At this stage, the values of F1, G1, F2, G2, A1, and B1 are undecided. However, in order to facilitate understanding, temporary values are assigned to F1, G1, F2, G2, A1, and B1 in FIG. The graph 1101, the graph 1102, the graph 1103, and the graph 1104 are illustrated using the graph.
In FIG. 11, the result occurrence index threshold value h used in the evaluation in FIG. 12 is also shown for reference.

FIG. 12 shows the relationship between the value of the result occurrence index P1 and the result occurrence index threshold h when the result R1 occurs, and the value of the result occurrence index P2 and the result occurrence index threshold h when the result R2 occurs. It is a figure showing a relation typically.
In FIG. 12, a graph 1201 indicates the time transition of the result occurrence index P2, and is the same graph as the graph 1104 described in FIG.
The coefficient value calculation unit 103 sets the uncertain coefficients F1, G1, F2, G2, A1 so that h ≦ Pi and the sum of the values of Pi−h (the sum of all event generation targets) is minimized. , B1 is obtained.
In FIG 12, P2 when the time of P1, and t _{= t q2} of t _{= t q1} is given below.
P1 = A ₁ (1-exp (−B ₁ (t _q1 −t ₁ ))) + F ₁ exp (−G ₁ (t _q1 −t _k1 )) + F ₂ exp (−G ₂ (t _q1 −t _k2 ) )
P2 = A ₁ (1-exp (−B ₁ (t _q2 −t ₂ ))) + F ₁ exp (−G ₁ (t _q2 −t _k3 ))
In this example, since the event generation targets belonging to the environment ET1 are only the event generation target O1 and the event generation target O2, the coefficient value calculation unit 103 performs h ≦ P1, h ≦ P2, and (P1−h) + (P2−h). ), The coefficient value calculation unit 103 obtains the values of the coefficients F1, G1, F2, G2, A1, and B1 (t ₁ , t _q1 , t _q2 , t _k1 , t _k2 , t _k3 and h are known).

As shown in FIG. 10, the event generation target O1 and the event generation target O2 have the same environment type and the same result type of the result event, and some of the cause events (C1, C3) The cause type is common.
As described above, when a plurality of event generation targets have the same environment type, the result type of the result event is the same, and the cause type of at least some of the cause events is the same, the coefficient value calculation unit 103 has a plurality of The event generation target calculation formulas are integrated as described above to perform the calculation.

Next, operation examples of the coefficient value calculation unit 103 and the result event candidate calculation unit 104 according to the present embodiment will be described using the specific values shown in FIGS.
For the sake of simplicity, the dates and times in FIGS. 3, 4, 5, 8, and 9 are written up to the date.

The coefficient value calculation unit 103 evaluates the date / time difference in terms of days.
Here, coefficient value calculation section 103 sets h = 1.
P1 when t = t _q1 and P2 when t = t _q2 are given below.
P1 = A ₁ (1-exp (−B ₁ (t _q1 −t ₁ ))) + F ₁ exp (−G ₁ (t _q1 −t _k1 )) + F ₂ exp (−G ₂ (t _q1 −t _k2 ) )
P2 = A ₁ (1-exp (−B ₁ (t _q2 −t ₂ ))) + F ₁ exp (−G ₁ (t _q2 −t _k3 ))
t _q1 -t ₁ = 70, t _q1 -t _k1 = 60, t _q1 -t _k2 = 40, t _q2 -t ₂ = 80, t _q2 -t _k3 = 65
Therefore, when a known value is substituted, the result is as follows.
P1 = A ₁ (1-exp (−70B ₁ )) + F ₁ exp (−60G ₁ ) + F ₂ exp (−40G ₂ )
P2 = A ₁ (1-exp (−80B ₁ )) + F ₁ exp (−65G ₁ )
From the above, the coefficient value calculation unit 103 sets the parameters A1, B1, which minimize (P1−h) + (P2−h) = P1 + P2-2 within the range where h = 1 ≦ P and h = 1 ≦ P2. The values of F1, F2, G1, and G2 are calculated.
The calculation results of A1, B1, F1, F2, G1, and G2 are as shown in FIGS.

The result event candidate calculation unit 104 uses the calculated values of the coefficients A1, B1, F1, F2, G1, and G2 to calculate the result occurrence index P3 for the new cause events C111 and C112 for the new event generation target O10. Find the formula.
The result event candidate calculation unit 104 obtains the mathematical expression of the result occurrence index P3 for each result type ID (here, only RT1 is described).
P3 (t) = A ₁ (1-exp (−B ₁ (t−t ₃ ))) + F ₁ exp (−G ₁ (t−t _k111 )) + F ₂ exp (−G ₂ (t−t _k112 ) )
= 1−exp (−0.0305 (t−t ₃ ))) + exp (−0.0377 (t−t _k111 )) + exp (−4.34 (t−t _k112 ))
If t ′ = t−t ₃ , then t−t _k111 = t′−45 and t−t _k111 = t′−85.
The result event candidate calculation unit 104 obtains t at which P3 (t) takes a maximum value.
When t ′ = 45, P3 takes a maximum value of 1.72.
The date corresponding to t ′ is 2015/05/17.
As illustrated in FIG. 9, the result event candidate calculation unit 104 sets an assumed period for determining that there is a possibility that a result event may occur because the value of the result occurrence index P3 exceeds the result occurrence index threshold value h. The result event candidate information 500 includes the start date and time and the expected occurrence period start date and time and end date and time.
The range where P3 (t) exceeds the result occurrence index threshold value h is 45 ≦ t ′ ≦ 235, and the corresponding date is 2015/05/16 to 2015/11/22.
As a result, the result event candidate information 500 shown in FIG. 9 is output.

*** Explanation of effects ***
As described above, the information processing apparatus 100 according to the present embodiment is based on the occurrence of a cause event, formulates the effect of decaying with time by the cause influence coefficient, and further determines the influence constantly received from the environment by the environment influence coefficient. Formulate.
Then, the information processing apparatus 100 calculates each coefficient from the state where each coefficient is indefinite so that the ratio at which the result occurrence index exceeds the threshold when the result event occurs is maximized.
The cause influence coefficient means the strength of the relationship between the cause event and the result event, and the environmental impact factor means the strength of the relationship between the environment and the result event.

According to the present embodiment, in order to calculate the value of the cause influence coefficient and the value of the environmental influence coefficient, the influence that decays with time from the cause event and the influence from the environment that always exists are quantified. be able to.
For this reason, the occurrence probability of the result event can be evaluated in consideration of the influence from the cause event and the influence from the environment.

As an example, in the case of evaluating maintenance staff education and technical improvement in maintenance service operations, the following effects can be achieved.
By making it possible to compare the influence of education and the influence of the external environment on the same scale index of the cause influence coefficient and the environment influence coefficient, the determination of whether to conduct multiple education or to improve the environment is quantified based on the same standard Can be compared.
For this reason, it is possible to easily make a comprehensive judgment in consideration of other quantitative parameters such as cost.
Even in long-term education planning, quality can be maintained and education costs can be minimized.
In addition, by setting priorities to measures in descending order of their impact on improving technical skills in education and environmental maintenance, and implementing measures in order of priority within limited resources, it is within the scope of actual constraints. Efficient technology improvement.

*** Explanation of hardware configuration ***
Finally, a hardware configuration example of the information processing apparatus 100 will be described with reference to FIG.
The information processing apparatus 100 is a computer.
The information processing apparatus 100 includes hardware such as a processor 901, an auxiliary storage device 902, a memory 903, a communication device 904, an input interface 905, and a display interface 906.
The processor 901 is connected to other hardware via the signal line 910, and controls these other hardware.
The input interface 905 is connected to the input device 907.
The display interface 906 is connected to the display 908.

The processor 901 is an IC (Integrated Circuit) that performs processing.
The processor 901 is, for example, a CPU (Central Processing Unit), a DSP (Digital Signal Processor), or a GPU (Graphics Processing Unit).
The auxiliary storage device 902 is, for example, a ROM (Read Only Memory), a flash memory, or an HDD (Hard Disk Drive).
The memory 903 is, for example, a RAM (Random Access Memory).
The information storage unit 101 illustrated in FIG. 1 is at least one of the auxiliary storage device 902 and the memory 903.
The communication device 904 includes a receiver 9041 that receives data and a transmitter 9042 that transmits data.
The communication device 904 is, for example, a communication chip or a NIC (Network Interface Card).
The input interface 905 is a port to which the cable 911 of the input device 907 is connected.
The input interface 905 is, for example, a USB (Universal Serial Bus) terminal.
The display interface 906 is a port to which the cable 912 of the display 908 is connected.
The display interface 906 is, for example, a USB terminal or an HDMI (registered trademark) (High Definition Multimedia Interface) terminal.
The input device 907 is, for example, a mouse, a keyboard, or a touch panel.
The display 908 is, for example, an LCD (Liquid Crystal Display).

The auxiliary storage device 902 includes a calculation formula generation unit 102, a coefficient value calculation unit 103, a result event candidate calculation unit 104 (hereinafter, a calculation formula generation unit 102, a coefficient value calculation unit 103, and a result event candidate calculation unit 104 shown in FIG. Are collectively stored as “parts”).
This program is loaded into the memory 903, read into the processor 901, and executed by the processor 901.
Further, the auxiliary storage device 902 also stores an OS (Operating System).
Then, at least a part of the OS is loaded into the memory 903, and the processor 901 executes a program that realizes the function of “unit” while executing the OS.
Although one processor 901 is illustrated in FIG. 14, the information processing apparatus 100 may include a plurality of processors 901.
A plurality of processors 901 may execute a program for realizing the function of “unit” in cooperation with each other.
In addition, information, data, signal values, and variable values indicating the processing results of “unit” are stored in the memory 903, the auxiliary storage device 902, or a register or cache memory in the processor 901.

The “part” may be provided as “circuitry”.
Further, “part” may be read as “circuit”, “process”, “procedure”, or “processing”.
“Circuit” and “Circuitry” include not only the processor 901 but also other types of processing circuits such as a logic IC or GA (Gate Array) or ASIC (Application Specific Integrated Circuit) or FPGA (Field-Programmable Gate Array). It is a concept to include.

DESCRIPTION OF SYMBOLS 100 Information processing apparatus, 101 Information storage part, 102 Calculation formula production | generation part, 103 Coefficient value calculation part, 104 Result event candidate calculation part, 200 Result generation | occurrence | production index threshold value, 300 Cause influence coefficient information, 400 Environmental influence coefficient information, 500 Result event Candidate information, 1011 cause event DB, 1012 result event DB, 1013 environment information DB, 1014 new cause event DB, 1021 environmental impact formulation unit, 1022 cause impact formulation unit, 1031 co-occurrence relation evaluation unit, 1032 mathematical programming calculation Department.

Claims (12)

1. The cause event occurrence time that is the time when the cause event occurred in the event occurrence target, the result event occurrence time that is the time when the result event occurred in the event occurrence target, and the environment to which the event occurrence target belongs is the event occurrence target. An information storage unit for storing event information in which an environmental impact time that is a time at which an influence starts is described;
   An environmental impact calculation formula that is a calculation formula for calculating a value of an environmental impact coefficient that is a saturation coefficient of an environmental impact that is an impact from the environment is generated using the result event occurrence time and the environmental impact time. , Using the cause event occurrence time and the result event occurrence time as a cause effect calculation formula that is a calculation formula for calculating a value of a cause influence coefficient that is an attenuation coefficient of a cause influence that is an influence from the cause event A calculation formula generation unit for generating
   A constraint condition and an objective function are set in the environmental impact calculation formula and the cause impact calculation formula, and the environmental impact calculation formula and the cause impact calculation formula in which the constraint condition and the objective function are set, An information processing apparatus comprising: a coefficient value calculation unit that calculates a value of an environmental influence coefficient and a value of the cause influence coefficient.
2. The coefficient value calculator is
   The information processing apparatus according to claim 1, wherein the value of the environmental influence coefficient and the value of the cause influence coefficient that minimize the objective function are calculated by mathematical programming.
3. The information processing apparatus further includes:
   The value of the environmental influence coefficient calculated by the coefficient value calculation section when a new cause event occurs after the value of the environmental influence coefficient and the value of the cause influence coefficient are calculated by the coefficient value calculation section. Is applied to the environmental influence calculation formula, and the value of the cause influence coefficient calculated by the coefficient value calculation unit is applied to the cause influence calculation formula to cause a result event that may be generated by the new cause event. The information processing apparatus according to claim 1, further comprising a result event candidate calculation unit that identifies a candidate.
4. The result event candidate calculation unit
   The information processing apparatus according to claim 3, wherein the occurrence probability of the result event candidate is calculated.
5. The result event candidate calculation unit
   The information processing apparatus according to claim 3, wherein a time when the result event candidate occurs is predicted.
6. The result event candidate calculation unit
   If there are multiple outcome event candidates, calculate the probability of occurrence for each outcome event candidate,
   The information processing apparatus according to claim 4, wherein the plurality of result event candidates are ranked in descending order of the calculated occurrence probability.
7. The calculation formula generation unit
   When a new result event occurs after the value of the environmental influence coefficient and the value of the cause influence coefficient are calculated by the coefficient value calculation unit, the occurrence time of the new result event is reflected to reflect the environmental influence Update the calculation formula and the causal effect calculation formula,
   The coefficient value calculator is
   The constraint condition and the objective function are set in the updated environmental impact calculation formula and the updated cause / effect calculation formula by the calculation formula generation unit, and the post-update is performed in which the constraint condition and the objective function are set. The information processing apparatus according to claim 1, wherein a new value of the environmental influence coefficient and a value of the cause influence coefficient are calculated from the environmental influence calculation formula and the updated cause influence calculation formula.
8. The calculation formula generation unit
   A value of an environmental impact reference coefficient that is a coefficient of a reference value of the intensity of environmental impact that is the environmental impact coefficient; and a value of an environmental impact saturation speed coefficient that is a coefficient of a speed at which the strength of the environmental impact is saturated Is generated using the result event occurrence time and the environmental impact time, and is a causal influence coefficient that is a coefficient of a reference value of the strength of the causal influence that is the causal influence coefficient. A cause effect calculation formula for calculating a value of a reference coefficient and a value of a cause effect attenuation rate coefficient that is a coefficient of a rate at which the intensity of the cause effect attenuates is expressed as the cause event occurrence time and the result event occurrence time. And using
   The coefficient value calculator is
   Based on the constraint condition, the objective function, the environmental impact time of the environmental impact calculation formula, the result event occurrence time, the cause event occurrence time and the result event occurrence time of the cause impact calculation formula, the environmental impact criteria The information processing apparatus according to claim 1, wherein a coefficient value, a value of the environmental influence saturation speed coefficient, a value of the cause influence reference coefficient, and a value of the cause influence attenuation speed coefficient are calculated.
9. The coefficient value calculator is
   The value of the environmental impact reference coefficient, the value of the environmental impact saturation rate coefficient, the value of the cause impact reference coefficient, and the value of the cause impact decay rate coefficient that minimize the value of the objective function by mathematical programming. The information processing apparatus according to claim 8 to calculate.
10. The calculation formula generation unit
    The information processing apparatus according to claim 1, wherein the environmental influence calculation formula and the cause influence calculation formula are generated by a combination of a polynomial function and an exponential function.
11. A cause event occurrence time that is a time when a cause event occurs in the event occurrence target, a result event occurrence time that is a time when a result event occurs in the event occurrence target, and an environment to which the event occurrence target belongs is the event Read event information from the storage area describing the environmental impact time, which is the time when it started to affect the occurrence target,
    The computer calculates an environmental impact calculation formula that is a calculation formula for calculating a value of an environmental impact coefficient that is a saturation coefficient of an environmental impact that is an impact from the environment, the result event occurrence time and the environmental impact time. The cause effect calculation formula, which is a calculation formula for calculating the value of the cause influence coefficient which is the attenuation coefficient of the cause influence which is the influence from the cause event, is generated using the cause event occurrence time and the result event occurrence Using time and
    The computer sets a constraint condition and an objective function in the environmental impact calculation formula and the cause impact calculation formula, and the environmental impact calculation formula and the cause impact calculation formula in which the constraint condition and the objective function are set. Information processing method for calculating the value of the environmental influence coefficient and the value of the cause influence coefficient.
12. The cause event occurrence time that is the time when the cause event occurred in the event occurrence target, the result event occurrence time that is the time when the result event occurred in the event occurrence target, and the environment to which the event occurrence target belongs is the event occurrence target. An information read process for reading event information from the storage area in which the environmental impact time, which is the time at which the impact began, is described;
    An environmental impact calculation formula that is a calculation formula for calculating a value of an environmental impact coefficient that is a saturation coefficient of an environmental impact that is an impact from the environment is generated using the result event occurrence time and the environmental impact time. , Using the cause event occurrence time and the result event occurrence time as a cause effect calculation formula that is a calculation formula for calculating a value of a cause influence coefficient that is an attenuation coefficient of a cause influence that is an influence from the cause event Calculation formula generation processing generated by
    A constraint condition and an objective function are set in the environmental impact calculation formula and the cause impact calculation formula, and the environmental impact calculation formula and the cause impact calculation formula in which the constraint condition and the objective function are set, An information processing program for causing a computer to execute a coefficient value calculation process for calculating an environmental influence coefficient value and a cause influence coefficient value.
    

Images