WO2013145584A1 - イベント相関検出システム - Google Patents
イベント相関検出システム Download PDFInfo
- Publication number
- WO2013145584A1 WO2013145584A1 PCT/JP2013/001481 JP2013001481W WO2013145584A1 WO 2013145584 A1 WO2013145584 A1 WO 2013145584A1 JP 2013001481 W JP2013001481 W JP 2013001481W WO 2013145584 A1 WO2013145584 A1 WO 2013145584A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- event
- probability
- event type
- minimum
- combination
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/027—Frames
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3466—Performance evaluation by tracing or monitoring
- G06F11/3476—Data logging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2201/00—Indexing scheme relating to error detection, to error correction, and to monitoring
- G06F2201/86—Event-based monitoring
Definitions
- the present invention relates to the technical field of information processing for detecting correlation of events.
- FIG. 21 is a block diagram showing a configuration of a general event correlation detection system.
- the event correlation detection system includes an event generation source 110, a window dividing unit 1001, an event type window table storage unit 1002, an event correlation rule engine unit 1003, a Graphical User Interface (hereinafter, “ 115 ”(abbreviated as“ GUI ”), event history database (hereinafter,“ database ”is abbreviated as“ DB ”) storage unit 1004, event type window division unit 1005, event type window table storage unit 1006, and rule generation unit 930. And an event correlation rule storage unit 1008 and the like.
- GUI Graphical User Interface
- DB event history database
- Patent Document 1 discloses an example of a rule generation method in the event correlation detection system shown in FIG.
- FIG. 22 is a diagram illustrating an example of a rule generation method in the event correlation detection system illustrated in FIG. Referring to FIG. 22, the rule generation method includes a mining kernel interface function 26, a mining kernel function 16, a mining result recording function 28 for recording the execution result of the mining kernel, and the like.
- the event history DB storage unit 1004 stores a history related to an event generated by the event generation source 110.
- FIG. 26 is a diagram conceptually illustrating information included in the event history DB.
- the event history DB storage unit 1004 includes an identifier for identifying an event (hereinafter abbreviated as “ID”) (E #), the time when the event occurred (the item “Time” shown in FIG. 26), and an attribute relating to the event type, etc. Are stored in association with each other.
- the event type window dividing unit 1005 divides the time recorded in the event history DB storage unit 1004 in time interval W [sec] increments. As a result, the event type window dividing unit 1005 classifies the event into a set of events (hereinafter referred to as “event type window”) according to the time when the event occurred.
- FIG. 23 is a diagram conceptually illustrating a state in which the occurrence of an event is classified into an event type window.
- the horizontal axis in FIG. 23 indicates the passage of time, and the time advances as it goes to the right.
- a triangle in FIG. 23 indicates that an event has occurred.
- events occur intermittently.
- the events classified by the event type window dividing unit 1005 are stored in an event window table as shown in FIG.
- FIG. 24 is a diagram showing a specific example of the event window table.
- the event window table stores W # and E # in association with each other for at least one event ID (E #) that occurred at a time related to the window ID (W #).
- FIG. 25 is a diagram showing an example of an image of the event type window table.
- the event type window table stores the event type window ID (W #) in association with the event type of the event belonging to the event type window.
- the rule generation unit 930 generates an event correlation rule while referring to a predetermined threshold given by the operator via the GUI 115.
- the event generation source 110 generates an event.
- the window dividing unit 1001 stores events that occur at a constant time interval W [sec] in the event type window table storage unit 1002.
- the event type window table storage unit 1002 further stores W # and the event type of the event that occurred in W # in association with each other.
- the event correlation rule engine unit 1003 displays correlated events on the GUI 115 according to the following processing while referring to the event correlation rule storage unit 1008. For example, when an event having an event type A and an event having an event type B occur, a rule that an event having an event type C occurs is expressed as an event correlation rule “A, B ⁇ C”. .
- the event correlation rule engine unit 1003 determines that the event E3 occurs if the event E1 and the event E2 occur. As a result, the event correlation rule engine unit 1003 creates an event correlation rule “A, B ⁇ C” by associating the event with the event type of the event.
- the mining kernel function 16 creates an event correlation rule according to the following process.
- the mining kernel interface function 26 has information on a predetermined minimum support level (or minimum support rate) and a predetermined minimum reliability level (or minimum reliability level), which are criteria for determination.
- the mining kernel function 16 uses a data access program or utility to read a history of events that have occurred in the past, and analyzes the relationship between events based on the read results. Next, the mining kernel function 16 records the correlation rule created by the analysis in the mining result recording function 28.
- k represents the number of types of event combinations held by the mining kernel function 16.
- Support level SR k ⁇ Nw, (However, it is assumed that a combination of events having k types of events appears in Nw windows in all event type windows).
- the mining kernel function 16 repeatedly calculates the support level SR while increasing k by 1 until the support level SR in all event type windows is less than the predetermined minimum support level.
- the number of combinations for which the mining kernel function 16 has completed the above calculation is represented by h.
- the mining kernel function 16 presents a combination of events having an event type having a support SR that is equal to or greater than a predetermined minimum support at (h ⁇ 1) time points.
- the mining kernel function 16 extracts an event that exists in the combination of event types presented as described above.
- the mining kernel function 16 calculates whether there is a correlation with the occurrence of the extracted event according to the following processing. It is assumed that the presented combination of event types is ⁇ A, B, C, D ⁇ .
- the mining kernel function 16 extracts one event type (denoted by “D”) from among them, and depends on the reliability defined by the following calculation formula (hereinafter referred to as “reliability TR”). , D is examined to see if it is correlated with the occurrence of A, B, or C.
- Reliability TR “number of event type windows in which ⁇ A, B, C, D ⁇ occurs” ⁇ “number of event type windows in which ⁇ A, B, C ⁇ occurs”
- the mining kernel function sets “A, B, C ⁇ D” as the event correlation rule if the reliability TR calculated as described above is equal to or greater than a predetermined minimum reliability.
- Patent Document 1 The problem in Patent Document 1 is that it is difficult to appropriately determine a predetermined minimum support level and a predetermined minimum reliability level.
- the mining kernel function 16 detects an erroneous result that event types that are not originally correlated are correlated if the predetermined minimum support value and the predetermined minimum reliability value are inappropriate. .
- the minimum support and the minimum reliability have been obtained through trial and error so as to obtain an acceptable error detection rate.
- the main object of the present invention is to provide an event correlation detection system or the like that can automatically set the minimum support level and the minimum reliability level so as to suppress the false detection rate to an acceptable level.
- an event correlation detection system is characterized by having the following configuration.
- the event correlation detection system is: For multiple event types, For each event type, an arrival rate calculation unit that calculates an arrival rate at which the event type arrives at a predetermined time interval; For each event type, an occurrence probability calculation unit that calculates a first probability that the event type occurs based on the arrival rate and the time interval; Threshold calculation for calculating a second probability that the plurality of event types occur simultaneously based on the first probability, and calculating a minimum support for the event type combination and a minimum reliability for the dependency rule based on the second probability And An event correlation rule engine for detecting correlation between the plurality of event types according to the minimum support level and the minimum reliability level.
- an event correlation detection method includes: For multiple event types, For each event type, calculate the arrival rate at which the event type arrives at a predetermined time interval; For each event type, calculate a first probability that the event type will occur based on the arrival rate and the time interval; Calculating a second probability that the plurality of event types occur simultaneously based on the first probability, calculating a minimum support for the event type combination and a minimum reliability for the dependency rule based on the second probability; A correlation between the plurality of event types is detected according to the minimum support level and the minimum reliability level.
- Another object of the present invention is to provide an event correlation detection apparatus having the above-described configuration, a computer program for realizing the corresponding method using a computer, and a computer-readable storage medium storing the computer program. It is also achieved by using it.
- the event correlation detection system according to the present invention can suppress the false detection rate to an acceptable level.
- 2 is a flowchart illustrating a procedure of processing executed by the event correlation detection system according to the first embodiment of the present invention. It is a block diagram showing the structure which the event correlation detection system which concerns on the 2nd Embodiment of this invention has. It is a flowchart which shows the procedure of the process which the event correlation detection system which concerns on the 2nd Embodiment of this invention performs. It is a flowchart which shows the procedure of the process which the event correlation detection system which concerns on the 2nd Embodiment of this invention performs.
- FIG. 1 is a block diagram showing a configuration of an event correlation detection system 101 according to the first embodiment of the present invention.
- the event correlation detection system 101 according to the first embodiment includes an arrival rate calculation unit 102, an occurrence probability calculation unit 103, a threshold value calculation unit 104, and an event correlation rule engine unit 160. .
- the arrival rate calculation unit 102 calculates the number of arrivals per unit time (hereinafter, this value is referred to as “arrival rate”) for each event.
- arrival rate the number of arrivals per unit time
- the occurrence probability calculation unit 103 is in a certain time interval W [sec (seconds)] (hereinafter also referred to as “predetermined time interval”) according to the arrival rate calculated by the arrival rate calculation unit 102.
- predetermined time interval the probability P (E
- the threshold value calculation unit 104 calculates the probability that all events included in the rule candidates will occur based on the probability P (E
- FIG. 2 is a flowchart showing a procedure of processing executed by the event correlation detection system 101 according to the first embodiment of the present invention.
- the arrival rate calculation unit 102 calculates the arrival rate for each event based on the history of occurrence of the event (step A10).
- the occurrence probability calculation unit 103 based on the arrival rate for each event calculated by the arrival rate calculation unit 102, the probability of occurrence of an event (step A20), the support SR and the reliability as described in the background art The degree TR is calculated (step A30).
- the threshold value calculation unit 104 calculates two types of probability distributions based on the probability calculated above and the support level SR, or based on the calculated probability and the reliability level TR (step A40). Next, the threshold calculation unit 104 calculates thresholds such that the area of the region corresponding to the false detection rate in the two types of probability distributions described above is equal to or less than the assumed false detection rate, and sets the calculated threshold to the minimum.
- the support level or the minimum reliability level is set (step A50).
- the event correlation rule engine unit 160 detects the correlation between the event types described above according to the minimum support value or the minimum reliability value calculated by the threshold value calculation unit 104 (step A60). .
- the event correlation detection system 101 calculates a threshold value that is equal to or lower than the assumed false detection rate, and calculates the calculated value as the minimum support level or the minimum reliability level. To do.
- the false detection rate can be suppressed to an acceptable level.
- FIG. 3 is a block diagram showing the configuration of the event correlation detection system 105 according to the second embodiment of the present invention.
- the event correlation detection system 105 includes a window division unit 140 that classifies events at regular time intervals W [sec], an event type window table storage unit 220 that stores information related to event type windows, and the like.
- the event correlation detection system 105 includes an event correlation rule engine unit 160 that detects correlated events based on values existing in the event correlation rule storage unit 170.
- the event correlation detection system 105 includes an arrival rate calculation unit 102 that calculates an arrival rate for each event from the value of the event history DB storage unit 130, and an event type at a certain time interval W [sec] from the event arrival rate. And an occurrence probability calculation unit 103 that calculates the probability of occurrence of the event having the event.
- the event correlation detection system 105 includes an event type window dividing unit 210 that classifies event types at regular time intervals W [sec] according to an event occurrence time in the event history DB storage unit 130, and an event correlation rule. And a rule generation unit 230 for generating.
- the rule generation unit 230 refers to the event type window stored in the event type window table storage unit 220 to generate an event group having a high coincidence rate as a rule candidate.
- the rule generation unit 230 stores the result in the rule candidate storage unit 240.
- the rule generation unit 230 generates an event correlation rule using the value stored in the rule candidate storage unit 240 and the threshold stored in the threshold storage unit 310.
- FIG. 5 is a block diagram showing the configuration of the rule generation unit 230 in the second embodiment of the present invention.
- the rule generation unit 230 includes a combination generation unit 410, a combination selection unit 450, a dependency rule generation unit 460, and a dependency rule selection unit 510.
- FIG. 6 is a block diagram showing the configuration of the rule candidate storage unit 240 in the second embodiment of the present invention.
- the rule candidate storage unit 240 includes an event combination storage unit 420 and a dependency rule storage unit 470.
- FIG. 7 is a block diagram showing the configuration of the threshold value calculation unit 104 according to the second embodiment of the present invention.
- the threshold value calculation unit 104 includes a minimum support level calculation unit 430 and a minimum reliability level calculation unit 480.
- FIG. 8 is a block diagram illustrating the configuration of the threshold storage unit 310 according to the second embodiment of the present invention. As illustrated in FIG. 8, the threshold storage unit 310 includes a minimum support level storage unit 440 and a minimum reliability level storage unit 490.
- FIG. 10 is a block diagram showing the configuration of the minimum reliability calculation unit 480 in the second embodiment of the present invention.
- the minimum reliability calculation unit 480 includes a probability distribution calculation unit 710, a probability distribution storage unit 720, and a distribution probability inverse calculation unit 730.
- FIG. 12 is a block diagram showing the configuration of the minimum support level calculation unit 430 in the second embodiment of the present invention.
- the minimum support level calculation unit 430 includes a probability distribution calculation unit 610, a probability distribution storage unit 620, and a distribution probability inverse calculation unit 630. The processing performed by each of these units will be described later together with the description of the flowchart.
- the event generation source (event generation source 110 in FIG. 21) generates an event.
- the event occurs according to the Poisson distribution.
- a data center can be cited as an example of an event generation source.
- the data center includes a large number of servers, network devices, storages, and the like. Many servers in a data center typically generate events independently of each other. When multiple devices generate events independently, such as servers in a data center, it can be assumed that the entire event occurs according to a Poisson distribution. If the event source is a data center, the user of this system is a data center operator.
- a GUI for example, an input device 2305 in a hardware configuration described later (FIG. 20) is an interface that allows a user to input an allowable false detection rate.
- the GUI for example, the output device 2304 in the hardware configuration (FIG. 20) described later
- the assumed false detection rate storage unit 120 stores a value representing a false detection rate acceptable by the user.
- the false detection rate is a probability that the event correlation rule stored in the event correlation rule storage unit 170 includes a rule having no correlation.
- FIG. 26 is a diagram conceptually illustrating information stored in the event history DB.
- the information stored in the event history DB in FIG. 26 is also used in the event correlation detection system 105 according to the second embodiment of the present invention.
- FIG. 3 is a block diagram showing the configuration of the event correlation detection system 105 according to the second exemplary embodiment of the present invention.
- the event history DB storage unit 130 stores a history related to events that have occurred in the past from the event generation source (event generation source 110 in FIG. 21).
- the event history DB stored in the event history DB storage unit 130 stores event occurrence times and event types in association with each other.
- the event history DB storage unit 130 may store the event history DB in association with values other than the parameters exemplified above.
- the window dividing unit 140 classifies events generated from event generation sources (event generation sources 110 in FIG. 21) into event type windows. Next, the window dividing unit 140 records information related to the classified event type window in the event type window table storage unit 220.
- FIG. 24 is a diagram showing a specific example of the event window table.
- the event window table is a table stored by the related event correlation detection system, and is also a table stored by the event correlation detection system in the second exemplary embodiment of the present invention.
- the event correlation rule engine unit 160 determines that there is a correlation in the occurrence of the event E1, the event E2, and the event E3 according to the event generation rule.
- the event correlation rule storage unit 170 stores a precondition composed of events having at least one event type and an outcome composed of events having a certain event type in association with each other. For example, when A, B, and C each represent an event type, the event correlation rule “A, B ⁇ C” indicates that “the event having the event type A and the event having the event B type occur as events occur. This represents a correlation that an event having type C occurs.
- the event correlation rule is “A, B ⁇ C”
- “Prerequisite” indicates “A, B” existing to the left of “ ⁇ ”
- “Consequence” exists to the right of “ ⁇ ”. Represents “C”.
- the event type window dividing unit 210 classifies the events stored in the event history DB storage unit 130 in units of event type windows. Next, the event type window dividing unit 210 associates an event type related to an event occurring in the event type window with an identifier representing the event type window (hereinafter abbreviated as “ID”), and the result is an event type. Saved in the window table storage unit 220.
- ID an identifier representing the event type window
- the event type window table storage unit 220 stores the event type for each event type window.
- FIG. 25 is a diagram showing an example of an image related to the event type window table.
- the event type window table associates an ID (W #) related to an event type window with a set of event types possessed by an event existing in the event type window.
- the event type window table is information that the related event correlation detection system has, and also information that the event correlation detection system in the second exemplary embodiment of the present invention has.
- the rule generation unit 230 generates an event correlation rule while referring to the threshold value, and stores the result in the event correlation rule storage unit 170.
- the configuration of the rule generation unit 230 will be described using an example as a subject.
- the rule generation unit 230 includes a combination generation unit 410, a combination selection unit 450, a dependency rule generation unit 460, and a dependency rule selection unit 510.
- A, B, C, D, and E exist as event types in the event type window.
- the combination selection unit 450 selects ⁇ A, B ⁇ from the event type combinations stored in the event combination storage unit 420.
- the combination generation unit 410 generates a new event type combination by adding one event type in the event type window for the event type combination ⁇ A, B ⁇ selected by the combination selection unit 450.
- the combination generation unit 410 creates event type combinations ⁇ A, B, C ⁇ , ⁇ A, B, D ⁇ , ⁇ A, B, E ⁇ , and stores the combinations in the event combination storage unit 420. Output.
- the combination selection unit 450 selects, from the event combinations stored in the event combination storage unit 420, event combinations related to the calculated support level SR greater than or equal to the minimum support level stored in the minimum support level storage unit 440.
- the combination selection unit 450 calculates the support s (A, B,%) Of the event type combinations A, B,.
- the dependency rule generation unit 460 generates a dependency rule according to the following process for at least one event combination selected by the combination selection unit 450. For example, it is assumed that the combination selection unit 450 selects the event type combination ⁇ A, B, C ⁇ . In that case, the dependency rule generation unit 460 generates “A, B ⁇ C”, “B, C ⁇ A”, and “C, A ⁇ B” as dependency rules from the event type combination, and the dependency rules are generated. Saved in the dependency rule storage unit 470.
- the dependency rule selection unit 510 selects a dependency rule having a reliability TR greater than or equal to the minimum reliability from the dependency rules generated by the dependency rule generation unit 460.
- the dependency rule selection unit 510 sets the selected dependency rule as an event correlation rule, and stores the result in the event correlation rule storage unit 170.
- the combination selection unit 450 calculates the reliability t (B, C ⁇ A) related to the dependency rule “B, C ⁇ A” according to Equation 2.
- T (B, C ⁇ A) (number of event type windows including A, B, C) ⁇ (number of event type windows including B, C) (Equation 2).
- the rule candidate storage unit 240 includes an event combination storage unit 420 and a dependency rule storage unit 470.
- the rule candidate storage unit 240 stores the work data temporarily created by the rule generation unit 230.
- the event combination storage unit 420 stores a combination of event types obtained as a result of combining a plurality of event types.
- the event type combinations include, for example, ⁇ A, B, C ⁇ , ⁇ A, B, D ⁇ , ⁇ A, B, E ⁇ and the like are assumed.
- the dependency rule storage unit 470 stores a rule that combines a “precondition” configured using a plurality of event types and a “consequence” configured using the event types.
- the arrival rate calculation unit 102 determines the event occurrence frequency for each event type (here, the event type is E) (here, the above-described event occurrence DB). In accordance with ( ⁇ E) [/ sec]). For example, when the event E occurs NE times in the time interval T, the arrival rate calculation unit 102 calculates the event occurrence frequency ⁇ E according to the process shown in Equation 3.
- FIG. 19 is a diagram illustrating an example of a reference table indicating the relationship between the event type and the arrival rate, which is stored in the event arrival rate table storage unit 260 according to the second embodiment of the present invention.
- the event arrival rate table storage unit 260 stores an event type and an arrival rate at which the event type arrives in association with each other.
- the occurrence probability calculation unit 103 calculates an occurrence probability according to the following process for each event type in which an event occurs once or more at a fixed time interval W.
- an event of event type E occurs according to a Poisson distribution.
- the occurrence probability calculation unit 103 represents an occurrence probability Pr (
- the occurrence probability table storage unit 280 stores the value calculated by the occurrence probability calculation unit 103 and the event type in association with each other as shown in Equation 4.
- FIG. 16 is a diagram illustrating an example of a reference table indicating the association between an event type and its occurrence probability, which is stored in the occurrence probability table storage unit according to the second embodiment of the present invention.
- the occurrence probability table storage unit 280 stores an event type and an occurrence probability Pr (
- the threshold value calculation unit 104 includes a minimum support level calculation unit 430 and a minimum reliability level calculation unit 480.
- the threshold calculation unit 104 calculates a threshold according to the following process, and stores the threshold in the threshold storage unit 310.
- the threshold value calculation unit 104 calculates the minimum support level and the minimum reliability level, and stores the calculated values in the minimum support level storage unit 440 and the minimum reliability level storage unit 490, respectively.
- the threshold calculation unit 104 calculates a threshold according to the following flow.
- the minimum support level calculation unit 430 includes a probability distribution calculation unit 610, a probability distribution storage unit 620, and a distribution probability inverse calculation unit 630.
- the minimum support level calculation unit 430 calculates a false detection rate according to the following process. As a result, the calculated false detection rate becomes a numerical value equal to or less than the assumed false detection rate stored in the assumed false detection rate storage unit 120.
- the processing method in the minimum support degree calculation unit 430 is as follows.
- the probability distribution calculation unit 610 refers to the event combination storage unit 420 and extracts a specific event type combination C from the event combinations.
- the event type combination C includes a plurality of event types.
- the probability distribution calculation unit 610 refers to the occurrence probability table storage unit 280 to extract the occurrence probability associated with the event type.
- the probability distribution calculation unit 610 performs the above-described processing for all event types included in the event type combination C.
- the probability distribution calculation unit 610 calculates the probability Pr (C
- W) ⁇ E ⁇ C_Pr (
- the probability distribution calculation unit 610 calculates the probability G (i) that the event type combination C occurs in i windows out of the Nw windows according to the process shown in Equation 6.
- G (i) C (Nw, i) P (C
- G (i) is equal to the binomial distribution B (Nw, Pr (C
- the probability distribution calculation unit 610 associates the generated window number i with the calculated G (i) and stores the result in the probability distribution storage unit 620.
- the probability distribution storage unit 620 stores the window number i and the value of G (i) in association with each other.
- FIG. 14 is a diagram illustrating characteristics of information stored in the probability distribution storage unit 620 according to the second embodiment of the present invention.
- the horizontal axis in FIG. 14 represents the number of windows (number of occurrences) in which the event combination C occurs.
- the vertical axis in FIG. 14 represents the probability (probability density) associated with the number of occurrence windows i described above in the probability distribution stored in the probability distribution storage unit 620.
- a curve indicated by a solid line in FIG. 14 represents a probability that an occurrence event type combination will occur when each event having an event type is assumed to occur independently.
- a curve indicated by a dotted line in FIG. 14 represents the occurrence probability when the occurrence of each event in the event type combination is not independent.
- the event type combination C includes an event type C1 and an event type C2.
- the probability distribution calculation unit 610 calculates the probability G (i) that an event type combination occurs on the assumption that an event having an event type C1 and an event having an event type C2 occur independently. To do.
- the solid line in FIG. 14 represents the probability G (i). That is, the solid line in FIG. 14 represents the binomial distribution B (Nw, Pr (C
- the apparatus when the minimum support degree is represented as s and the occurrence frequency is higher than (s ⁇ Nw) shown on the vertical axis, the apparatus according to the present embodiment causes each event in the event type combination to generate each other. Is determined to be dependent. However, the events B and C in the event type combination are independent of each other. Therefore, in FIG. 14, the portion shown in a lattice shape indicates that an erroneous result is detected.
- the distribution probability inverse calculation unit 630 calculates the minimum s satisfying Expression 7.
- This s may be calculated by substituting the value of i sequentially from a large value to find a value that does not satisfy Equation 7, or by analytically calculating it using a method such as the Newton method. Good.
- the method for calculating the minimum s is not limited to the method exemplified above.
- the dotted line shown in FIG. 14 represents the probability distribution when the occurrence of each event B, C in the event type combination is not independent.
- the area surrounded by the dotted line in FIG. 14 and the right part of the vertical line where the occurrence frequency shown on the horizontal axis indicates (s ⁇ Nw) is shown in FIG. 14 by the apparatus according to the present embodiment for each event B, C in the event type combination. If the occurrence of is not independent, it is the part that detects correctly.
- a region surrounded by a dotted line in FIG. 14 and a left portion of a vertical line in which the occurrence frequency shown on the horizontal axis represents (s ⁇ Nw) (shaded portion shown in FIG. 14) is below the minimum support level. Therefore, the device according to the present embodiment does not select this part.
- the minimum support level calculation unit 430 calculates the minimum support level by calculating s that satisfies Expression 7. As a result, the present embodiment according to the present invention can suppress the false detection rate to an acceptable level.
- the minimum reliability calculation unit 480 calculates the minimum reliability according to the process described below, and stores the calculated value in the minimum reliability storage unit 490. In the present embodiment, when such processing is performed, the value of the false detection rate becomes equal to or less than the value of the assumed false detection rate stored in the assumed false detection rate storage unit 120.
- the minimum reliability calculation unit 480 includes a probability distribution calculation unit 710, a probability distribution storage unit 720, and a distribution probability inverse calculation unit 730.
- the probability distribution calculation unit 710 refers to the value associated with the occurrence probability table storage unit 280 and the dependency rule R stored in the dependency rule storage unit 470.
- the probability distribution calculation unit 710 calculates the probability Pr (R) that the dependency rule is generated, assuming that the event having the event type included in the dependency rule R occurs independently. For example, when the dependency rule R is a dependency rule “B, C,... ⁇ A”, the probability distribution calculation unit 710 calculates the probability Pr (R
- W) Pr (
- the probability distribution calculation unit 710 calculates the probability distribution according to the binomial distribution B (Nw, Pr (R
- Nw represents the number of windows stored in the event history DB.
- the probability distribution calculation unit 710 stores the calculated probability distribution in the probability distribution storage unit 720.
- FIG. 15 is a diagram illustrating the characteristics of the probability distribution storage unit 720 according to the second embodiment of the present invention.
- the horizontal axis of FIG. 15 represents the number of windows (number of occurrences) in which the dependency rule has occurred.
- the vertical axis in FIG. 15 represents the probability density stored in the probability distribution storage unit 720.
- the solid line in FIG. 15 represents the probability that an occurrence event type combination will occur when each event having an event type is assumed to occur independently.
- the dotted line in FIG. 15 represents the probability when the occurrence of each event in the event type combination is not independent.
- the dependency rule R stored in the dependency rule storage unit 470 is assumed to be “B ⁇ A”.
- the solid line in FIG. 15 represents the probability distribution when an event having event type A and an event having event type B occur independently.
- the probability distribution calculation unit 710 calculates the probability distribution as a binomial distribution B (Nw, Pr (R
- the apparatus according to the present embodiment selects a part having a higher occurrence frequency than t ⁇ Nw as an event in which events occur depending on each other.
- the solid line in FIG. 15 represents a value calculated by the probability distribution calculation unit 710 on the assumption that an event having an event type A and an event having an event type B occur independently. Therefore, the distribution probability inverse calculation unit 730 is an area surrounded by the solid line in FIG. 15 and the right side of the vertical line where the occurrence frequency shown on the horizontal axis is (t ⁇ Nw) (the part shown in a grid pattern in FIG. 15). ) Is detected as an event having a dependency relationship with each other. That is, the detection is an incorrect result.
- the distribution probability inverse calculation unit 730 calculates the minimum t that satisfies Equation 9 when the assumed erroneous detection rate storage unit 120 is set to p0.
- Equation 9 (Where dR (Nw, W, R) (i) is the probability of taking the value i for the probability distribution dR (Nw, W, R), ( ⁇ i> t_Nw ⁇ dR (Nw, W, R) (i)) calculates the sum of Nw ⁇ dR (Nw, W, R) (i) for i having a value larger than t. i represents a natural number).
- the distribution probability inverse calculation unit 730 calculates the minimum reliability t according to the processing method described above with reference to the values of the probability distribution storage unit 720 and the assumed false detection rate storage unit 120.
- This t can be calculated by substituting the value of i sequentially from a large value and calculating a value that does not satisfy Equation 9, or by calculating it analytically and using a method such as the Newton method. Also good.
- the method for calculating the minimum t is not limited to the method exemplified above.
- the threshold value calculation unit 104 calculates a threshold value according to the following process, and stores the calculated value in the threshold value storage unit 310.
- the rule generation unit 230 refers to the threshold stored in the threshold storage unit 310.
- the threshold storage unit 310 includes a minimum support level storage unit 440 and a minimum reliability level storage unit 490.
- the minimum support level storage unit 440 stores a threshold value that the rule generation unit 230 refers to.
- the threshold calculated by the minimum support level calculation unit 430 represents a lower limit value of the support level SR regarding the event combination.
- the minimum reliability storage unit 490 stores a threshold value referred to by the rule generation unit 230.
- the threshold calculated by the minimum reliability calculation unit 480 represents a lower limit value in the reliability TR regarding the dependency rule.
- FIG. 9 is a flowchart regarding event correlation detection processing according to the second embodiment of the present invention.
- the window dividing unit 140 receives an event generated by an event generation source (event generation source 110 in FIG. 21). Then, the window dividing unit 140 classifies the events received at regular time intervals W into event type windows, and sets the event type window ID (W #) and the event type set of the events belonging to the event type window. Are recorded in the event type window table storage unit 220 (step E110).
- the event correlation rule engine unit 160 selects a correlated event among the events recorded in the event type window table storage unit 220 according to the value of the event correlation rule storage unit 170 (step E120). .
- FIGS. 3 to 8 and FIG. FIG. 4A and FIG. 4B are flowcharts showing a procedure of processing executed by the event correlation detection system 105 according to the second exemplary embodiment of the present invention.
- the arrival rate calculation unit 102 calculates the arrival rate for each event existing in the event history DB storage unit 130 according to the above-described process while referring to the value of the event history DB storage unit 130.
- the arrival rate calculation unit 102 associates the event with the calculated arrival rate at which the event arrives, and records it in the event arrival rate table storage unit 260 (step A210 in FIG. 4A).
- the occurrence probability calculation unit 103 refers to the value of the event arrival rate table storage unit 260 and then calculates the event occurrence probability at a fixed time interval W [sec] for each event existing in the event arrival rate table storage unit 260. The calculation is performed according to the processing as described above. Then, the occurrence probability calculation unit 103 stores the calculated value in the occurrence probability table storage unit 280 (step A220).
- the event type window dividing unit 210 classifies events as event type windows in units of a fixed time interval W [sec] based on the values stored in the event history DB storage unit 130.
- the event type window dividing unit 210 stores the classified event type window in the event type window table storage unit 220 in association with the event type and the window (step A110 in FIG. 4A).
- the process flow from step A210 to step A220 and the process of step A110 may be performed in parallel, or one of the two may be performed first. Note that step A210, step A220, and step A110 are the same as the processing in FIG.
- step B120 the rule generation unit 230 sets the number of event combinations as 2 (step B120).
- the combination generation unit 410 generates an event combination as described above for the number of event combinations set by the rule generation unit 230 based on the value of the event type window table storage unit 220.
- the rule generation unit 230 stores the created event combination in the event combination storage unit 420 (step B130).
- the combination generation unit 410 combines two types of event types when the rule generation unit 230 sets the number of event combinations to two.
- the combination generation unit 410 creates an event combination by combining all event types.
- the rule generation unit 230 increases the number of event combinations by one.
- the combination generation unit 410 creates an event combination according to the number of event combinations set by the rule generation unit 230 and stores the created event combination in the event combination storage unit 420.
- the minimum support level calculation unit 430 refers to the value of the occurrence probability table storage unit 280 and the value of the assumed false detection rate storage unit 120 for each event combination stored in the event combination storage unit 420, According to the processing method shown in Equation 9, the minimum support level is calculated, and the calculated value is stored in the minimum support level storage unit 440 (step B140).
- the combination selection unit 450 calculates the support level SR in the event combination.
- the combination selection unit 450 compares the calculated support level SR with the value existing in the minimum support level storage unit 440. Thereafter, the combination selection unit 450 leaves only event combinations having a large support level SR calculated from the minimum support level value (step B150). If the combination selection unit 450 determines that there is a combination in which the calculated support level SR is larger than the value present in the minimum support level storage unit 440 (YES in step B160), the above-described support is performed. Event combinations having a large degree SR are temporarily stored in the combination selection unit 450.
- the combination selection unit 450 increases the event combination by one (step B170), adds one event to the combination of the temporarily stored events in the combination selection unit 450, and then performs step Return to B130.
- the combination selection unit 450 determines that each calculated support level SR is smaller than the value existing in the minimum support level storage unit 440 (determined as NO in Step B160), the combination selection unit 450. , The dependency rule generation unit 460 generates a dependency rule (step B210).
- the minimum reliability calculation unit 480 refers to the value existing in the occurrence probability table storage unit 280 and the value existing in the assumed false detection rate storage unit 120 for each dependency rule generated by the dependency rule generation unit 460. However, the minimum reliability is calculated, and the calculated value is stored in the minimum reliability storage unit 490 (step B220).
- the dependency rule selection unit 510 calculates the reliability TR for each dependency rule generated by the dependency rule generation unit 460.
- the dependency rule selection unit 510 compares the calculated reliability TR with the value existing in the minimum reliability storage unit 490 associated with the dependency rule. Then, the dependency rule selection unit 510 leaves only the dependency rule having a higher value of the reliability TR calculated than the value of the minimum reliability from the dependency rules generated by the dependency rule generation unit 460 (step B230).
- the dependency rule selection unit 510 generates the dependency rule extracted as described above as an event correlation rule (step B240), and stores the created event correlation rule in the event correlation rule storage unit 170.
- FIG. 12 is a block diagram showing the configuration of the minimum support level calculation unit in the second embodiment of the present invention.
- FIG. 13 is a flowchart relating to processing performed by the minimum support level calculation unit according to the second embodiment of the present invention.
- dC (Nw, W, C) B (Nw, ⁇ E ⁇ C_Pr (
- ⁇ i_C (Nw, i) ⁇ P ⁇ i ⁇ (1-P) ⁇ i means taking the sum of C (Nw, i) ⁇ P ⁇ i ⁇ (1-P) ⁇ i for i).
- the distribution probability inverse calculation unit 630 generates “Pr (X> s ⁇ Nw
- the maximum s satisfying (Nw, W, C)) ⁇ the assumed erroneous detection rate p0 ” is calculated, and the calculated value is stored in the minimum support degree storage unit 440 (step C120).
- dC (Nw, W, C)) represents the probability that the random variable X takes a value larger than s ⁇ Nw in the probability distribution represented by dC (Nw, W, C).
- FIG. 10 is a block diagram showing the configuration of the minimum reliability calculation unit 480 in the second embodiment of the present invention.
- FIG. 11 is a flowchart relating to processing performed by the minimum reliability calculation unit 480 in the second embodiment of the present invention.
- the probability distribution calculation unit 710 determines that the event A is an event B, an event C,. And a probability distribution dR (Nw, W, R) is calculated.
- the calculation method is a method of calculating a binomial distribution B (Nw, Pr (A
- the probability distribution calculation unit 710 stores the calculated value in the probability distribution storage unit 720 (step D110).
- the dependency rule does not necessarily have to be “B, C,... ⁇ A” as described above.
- the distribution probability inverse calculation unit 730 performs “in the probability distribution dR (Nw, W, R) calculated by the probability distribution calculation unit 710 for each dependency rule“ B, C,...
- dR (Nw, W, R)) ⁇ assumed false detection rate p0 ” is calculated.
- the distribution probability inverse calculation unit 730 stores the calculated value in the minimum reliability storage unit 490 (step C130).
- the distribution probability inverse calculation unit 730 is the minimum that satisfies the condition that the probability that X> t is less than or equal to p0 in the probability distribution dR (Nw, W, R) described above. T is calculated as the minimum reliability.
- the event occurrence distribution probability is assumed to be a Poisson distribution.
- Each event is assumed to occur probabilistically independently.
- the calculation is performed according to the calculation method as described above.
- the threshold values calculated by the present embodiment that is, the minimum support level and the minimum reliability level
- the threshold values calculated by the present embodiment can be set so as not to exceed an allowable false detection rate.
- FIG. 26 is a diagram conceptually illustrating information stored in the event history DB storage unit 130.
- the event history DB also uses the event correlation detection system in the second embodiment of the present invention.
- the table existing in the event history DB storage unit 130 associates the event ID (E #), the time when the event occurred (TIME), the event type, and the like.
- the table existing in the event history DB storage unit 130 may be associated with an item other than the above, or does not necessarily include the above item.
- the event types associated with each E # are four types of events, namely “QueryError” and “QueryError”.
- DBError “ NWCongestion ”, and“ TooManyRequest ”.
- the arrival rate calculation unit 102 refers to a value existing in the event history DB storage unit 130 and extracts one event from the value.
- the arrival rate calculation unit 102 calculates the arrival rate for the extracted event according to the method described above, associates the calculated value with the event type, and stores it in the event arrival rate table storage unit 260.
- the arrival rate calculation unit 102 performs such processing for all events existing in the event history DB storage unit 130 (step A210 in FIG. 4A).
- FIG. 19 is a diagram illustrating an example of a reference table indicating the relationship between the event type and the arrival rate, which is stored in the event arrival rate table storage unit 260 according to the second embodiment of the present invention.
- the event arrival rate table associates the event type with the calculated arrival rate.
- the event having the QueryError type is 3 hours and 20 minutes from the time “2011/11/23 10:00: 00” to the time “2011/11/23 13:20”, that is, 12,000 seconds.
- the occurrence probability calculation unit 103 calculates the event occurrence probability at a fixed time interval W [sec] for each event as follows with reference to the event arrival rate table storage unit 260, and the calculated value Is stored in the occurrence probability table storage unit 280 (step A220).
- the occurrence probability calculation unit 103 determines the probability that a QueryError type event will occur according to Equation 4, Pr (
- the occurrence probability calculation unit 103 associates the calculated value with its event type, and stores them in the occurrence probability table storage unit 280.
- FIG. 16 is a diagram illustrating an example of a reference table indicating the relationship between event types and occurrence probabilities stored in the occurrence probability table storage unit 280 according to the second embodiment of the present invention.
- the event type window dividing unit 210 divides the time at a constant time interval W [sec]. Here, each divided time is called a window.
- the event type window dividing unit 210 searches the event history DB storage unit 130 for an event that has occurred in the window.
- the event type window dividing unit 210 associates the event type of the event with the window and stores it in the event type window table storage unit 220 (step A110 in FIG. 4A).
- the rule generation unit 230 operates as follows. After step A220 and step A110 are completed, the rule generation unit 230 sets the initial event combination to 2 (step B120). Next, the combination generation unit 410 refers to the value of the event type window table storage unit 220 and generates event combinations for the number of event combinations set by the rule generation unit 230 (step B130). That is, in the initial stage, the combination generation unit 410 combines two types of events.
- the combination generation unit 410 generates the following six types of event combinations.
- the minimum support degree calculation unit 430 refers to a table in the occurrence probability table storage unit 280 and a value existing in the assumed false detection rate storage unit 120 for each event combination, and a processing method as exemplified below Accordingly, the minimum support level storage unit 440 is calculated (step B140).
- the minimum support level calculation unit 430 performs processing as follows. However, in this example, it is assumed that the assumed erroneous detection rate storage unit 120 stores a value of 0.1. Hereinafter, the operation of the minimum support level calculation unit 430 will be described by taking as an example the case of calculating the minimum support level in an event type combination of an event having a QueryError type and an event having a DBError type.
- the minimum support level calculation unit 430 performs the calculation assuming that an event having the QueryError type and an event having the DBError type occur probabilistically independently. That is, the minimum support level calculation unit 430 calculates the probability that the above events occur simultaneously in the window at a constant time interval W [sec] according to the following process.
- the minimum support degree calculation unit 430 calculates min ⁇ s
- the calculated value is set as the minimum support level.
- X in the above equation is a random variable representing the number of windows in which QueryError and DBError occur simultaneously. Further, the above equation means that the minimum s satisfying Pr (X> Nws
- the combination selection unit 450 compares the calculated support level SR of each event combination with the minimum support level value associated with the combination. Thereafter, the combination selection unit 450 leaves only event combinations having a large support level SR calculated from the minimum support level value (step B150).
- FIG. 17 shows an example of a table that associates the combination of the two types of events as described above, the support SR calculated by the combination, and the minimum support.
- the table of FIG. 17 represents the support level SR calculated for the above six event type combinations.
- the rule generation unit 230 recognizes that there is an event type combination having a support level SR larger than the minimum support level value.
- the event type combinations having a greater support level SR calculated than the minimum support level value are only combinations of the QueryError type and the TooMany Request type.
- the values of the support level SR in the other event type combinations are all smaller than the value of the minimum support level. Therefore, the combination selection unit 450 extracts an event type combination in which the QueryError type and the TooManyRequest type are combined.
- the rule generation unit 230 increases the length of the event type combination by 1 (step B170). That is, in this example, since the length of the event type combination is 3, the combination generation unit 410 creates three event combinations according to 3 set by the rule generation unit 230, and the result is the event combination storage unit. Save to 420. In this example, the event combination storage unit 420 creates the following event type combinations for the event type combinations as described above.
- the minimum support level calculation unit 430 calculates a minimum support level for each of these two types of event combinations according to the method described above (step B149).
- the combination selection unit 450 calculates the support level SR of the event type combination, and compares the calculated value of the support level SR with the value existing in the minimum support level storage unit 440. As described above, the combination selection unit 450 leaves only the event type combinations having a greater support level SR calculated from the minimum support level value (step B150).
- an event type combination of an event having a QueryError type and an event having a TooMany Request type, in which the calculated support level SR has finally exceeded the minimum support level, is processed in the dependency rule generation unit 460. Become a target.
- the dependency rule generation unit 460 reads the values “QueryError” and “TooManRequest” by referring to the values existing in the event combination storage unit 420. Next, the dependency rule generation unit 460 performs the following process to generate a dependency rule (step B210).
- the combination selection unit 450 leaves the event type combination of the event having the QueryError type and the event having the TooManyRequest type, and the dependency rules are of the following two types.
- the minimum reliability calculation unit 480 calculates the minimum reliability by performing the following processing while referring to the value in the occurrence probability table storage unit 280 and the value in the assumed false detection rate storage unit 120. The calculated value is stored in the minimum reliability storage unit 490 (step B220).
- the number of times an event having a QueryError type occurs is 32 times. Further, it is assumed that the number of windows in which an event having a QueryError type and an event having a TooManyRequest type are generated is 21 times.
- the minimum reliability calculation unit 480 calculates Pr (TooManRequest
- the dependency rule selection unit 510 calculates the reliability TR for each dependency rule. Thereafter, the dependency rule selection unit 510 compares the calculated reliability TR value with the value associated with the dependency rule in the minimum reliability storage unit 490. Based on the comparison result, the dependency rule selection unit 510 leaves only the dependency rule having a high reliability TR calculated from the minimum reliability value (step B230).
- FIG. 18 is a diagram illustrating an example of the minimum reliability and the reliability TR in the second embodiment of the present invention.
- the leftmost column in FIG. 18 represents A in the dependency rule “A ⁇ B”, and the uppermost row in FIG. 18 represents B in the dependency rule “A ⁇ B”.
- the numerical value in the region where the row indicating A and the column indicating B intersect represents the reliability TR value of the calculated dependency rule “A ⁇ B”.
- the reliability TR of the dependency rule “QueryError ⁇ TooManRequest” is 0.65625
- the reliability TR of the dependency rule “TooManRequest ⁇ QueryError” is 0.84.
- the dependency rule selection unit 510 sets both of the two dependency rules as event correlation rules according to the result of the comparison as described above.
- the dependency rule selection unit 510 stores the two dependency rules in the event correlation rule storage unit 170 (step B240).
- FIG. 9 is a flowchart relating to event correlation detection processing according to the second embodiment of the present invention. Next, the event correlation detection process will be described using a specific example with reference to FIG.
- the window dividing unit 140 receives an event generated by an event generation source (for example, the event generation source 110 in FIG. 21). Next, the window dividing unit 140 classifies the events at regular time intervals W according to the time of occurrence. Then, the window dividing unit 140 associates the divided window with an event that has occurred in the window, and records the result in the event type window table storage unit 220 (step E110).
- an event generation source for example, the event generation source 110 in FIG. 21.
- the window dividing unit 140 classifies the events at regular time intervals W according to the time of occurrence. Then, the window dividing unit 140 associates the divided window with an event that has occurred in the window, and records the result in the event type window table storage unit 220 (step E110).
- FIG. 24 is a diagram showing a specific example of the event window table.
- the false detection rate can be suppressed to an acceptable level.
- FIG. 20 is a block diagram schematically showing a hardware configuration of a calculation processing device capable of realizing the event correlation detection system according to each embodiment of the present invention.
- an event correlation detection system may be realized using at least two types of calculation processing devices physically or functionally.
- the event correlation detection system may be realized as a dedicated device.
- FIG. 20 is a diagram schematically showing a hardware configuration of a calculation processing apparatus capable of realizing the event correlation detection system according to the first or second embodiment.
- the calculation processing device 2306 includes a CPU (Central Processing Unit) 2301, a memory 2302, a disk 2303, an output device 2304, an input device 2305, and a nonvolatile recording medium 2307 (hereinafter also referred to as “recording medium”).
- CPU Central Processing Unit
- the non-volatile recording medium 2307 can be read by a computer, such as a compact disc (Compact Disc), a digital versatile disc (Digital Versatile Disc), a Blu-ray Disc (Blu-ray Disc), a universal serial bus memory (USB memory), and the like. It points to the program and keeps it portable even without power.
- the nonvolatile recording medium 2307 is not limited to the above-described medium. Further, the program may be carried via a communication network instead of the non-volatile recording medium 2307.
- the distribution probability inverse calculation unit 730 copies a software program (computer program) stored in the disk 2303 to the memory 2302 at the time of execution, and executes arithmetic processing.
- the distribution probability inverse calculation unit 730 reads data necessary for program execution from the memory 2302. When display is necessary, the distribution probability inverse calculation unit 730 displays the output result on the output device 2304. When a program is input from the outside, the distribution probability inverse calculation unit 730 reads the program from the input device 2305.
- the distribution probability inverse calculation unit 730 interprets and executes the event correlation detection program stored in the memory 2302.
- the distribution probability inverse calculation unit 730 sequentially performs processes according to the flowcharts (FIGS. 2, 4A, 4B, 9, 11, and 13) and expressions referred to in the above-described embodiments.
- the present invention can also be realized by such an event correlation detection program. Furthermore, it can be understood that the present invention can also be realized by a computer-readable recording medium in which the event correlation detection program is recorded.
- an arrival rate calculation unit that calculates an arrival rate at which the event type arrives at a predetermined time interval
- an occurrence probability calculation unit that calculates a first probability that the event type occurs based on the arrival rate and the predetermined time interval
- a threshold calculation unit that calculates a second probability that the plurality of event types occur based on the first probability, and calculates a minimum support for the event type combination and a minimum reliability for the dependency rule based on the second probability
- An event correlation detection system comprising: an event correlation rule engine that detects correlation between the plurality of event types according to the minimum support level and the minimum reliability level.
- Appendix 2 Further comprising a window dividing unit that classifies the event types for each predetermined time interval, associates each predetermined time interval with the classified set of event types, and generates a window;
- the event correlation detection system according to attachment 1 wherein the arrival rate calculation unit performs calculation according to the window.
- a combination generation unit that generates a first combination by combining the plurality of event types existing in the window and collecting the sets;
- a support level that is a ratio of the number of the windows in which all event types existing in the set included in the first combination appear and the number of all the windows is calculated, and the support level is calculated from the first combination.
- a combination selection unit that selects the specific combination having a value larger than the minimum support as a second combination;
- a dependency rule generating unit that generates a dependency rule that assumes one event type as a result and a remaining event type as a precondition among the event types in the pair existing in the second combination;
- a dependency rule selection unit that calculates a reliability that is a ratio of the number of windows in which the event type appears in the dependency rule appears and the number of windows in which the precondition appears.
- the event correlation detection system according to supplementary note 2, wherein the event correlation rule engine unit detects a correlation between the event types based on the magnitude of the reliability and the minimum reliability.
- the combination generation unit sets the number of combinations to 2, The combination generation unit performs a first process of creating the set by combining the event types for the number of combinations.
- the event correlation rule engine unit performs a second process of holding the second combination as a third combination when it is determined that there is a set having the reliability higher than the minimum reliability. Later Performing a third process of increasing the number of combinations by one;
- the event correlation rule engine unit sequentially repeats the first process to the third process until it is determined that there is no set whose reliability is greater than the minimum reliability.
- the event correlation rule engine unit determines that there is no group having the reliability higher than the minimum reliability, the event correlation rule engine unit last holds the specific The event correlation detection system according to appendix 3, which outputs three combinations.
- the threshold calculation unit includes: A probability distribution calculation unit that calculates the second probability based on the first probability and generates a probability distribution G (i) by combining the calculated values; In the probability distribution G (i) (where i is a natural number), the sum of G (i) when the value of i is equal to or greater than a specific value is calculated, and the product of the calculated sum and the total number of windows is a predetermined assumption error.
- the event correlation detection system according to appendix 2, further comprising: a distribution probability calculation unit that outputs the minimum specific value in a range that is equal to or lower than a detection rate.
- the threshold calculation unit includes: Pr (
- W) representing the first probability is calculated
- the second probability is calculated based on the first probability
- the threshold calculation unit includes: Pr (R
- W) Pr (
- the threshold value calculation unit calculates the probability distribution according to a binomial distribution calculation process of the number of time intervals and the first probability existing in the window, and calculates the minimum support based on the probability distribution.
- the event correlation detection system according to attachment 5.
- the threshold calculation unit calculates the probability distribution according to a binomial distribution calculation process of the number of time intervals and the first probability existing in the window, and calculates the minimum reliability based on the probability distribution.
- the event correlation detection system according to attachment 5.
- the threshold calculation unit includes: ( ⁇ i> s_Nw ⁇ G (i)) ⁇ p0, (Where Nw is the number of all the windows, G (i) is the probability distribution calculated based on ⁇ E, ( ⁇ i> s_Nw ⁇ G (i)) is the sum of Nw ⁇ G (i) for i greater than t, p0 is the assumed false detection rate, i is a natural number),
- the event correlation detection system according to supplementary note 5, wherein a minimum s satisfying the above is calculated and the calculated value is the minimum support level.
- the threshold calculation unit includes: ( ⁇ i> t_Nw ⁇ dR (Nw, W, R) (i)) ⁇ p0 (Where Nw is the number of all the windows, dR (Nw, W, R) (i) is the probability distribution calculated based on the dependency rule R; ( ⁇ i> t_Nw ⁇ dR (Nw, W, R) (i)) is the sum of Nw ⁇ dR (Nw, W, R) (i) for i larger than t, p0 is the assumed false detection rate, i is a natural number),
- the event correlation detection system according to appendix 5, wherein the minimum t satisfying the above is calculated and the calculated value is the minimum reliability.
Abstract
Description
(ただし、k種類イベント型を持つイベントの組み合わせが、全イベント型ウィンドウ中、Nw個のウィンドウに現れるとする)。
マイニングカーネル機能は、上記のように計算した信頼度TRが、所定の最小信頼度以上ならば「A,B,C⇒D」をイベント相関規則とする。
複数のイベント型を対象として、
前記イベント型ごとに、前記イベント型が所定の時間間隔において到着する到着率を算出する到着率計算部と、
前記イベント型ごとに、前記到着率と前記時間間隔とに基づいて、前記イベント型が発生する第1確率を算出する発生確率計算部と、
前記第1確率に基づいて、前記複数のイベント型が同時に発生する第2確率を計算し、前記第2確率に基づいてイベント型組み合わせに対する最小支持度及び依存ルールに対する最小信頼度を算出する閾値計算部と、
前記最小支持度と前記最小信頼度に応じて前記複数のイベント型間における相関を検出するイベント相関用ルールエンジン部とを
備えることを特徴とする。
複数のイベント型を対象として、
前記イベント型ごとに、前記イベント型が所定の時間間隔において到着する到着率を計算し、
前記イベント型ごとに、前記到着率と前記時間間隔とに基づいて、前記イベント型が発生する第1確率を計算し、
前記第1確率に基づいて、前記複数のイベント型が同時に発生する第2確率を計算し、前記第2確率に基づいてイベント型組み合わせに対する最小支持度及び依存ルールに対する最小信頼度を計算し、
前記最小支持度と前記最小信頼度に応じて前記複数のイベント型間における相関を検出することを特徴とする。
図1は、本発明の第1の実施形態に係るイベント相関検出システム101が有する構成を示すブロック図である。図1を参照すると、第1の実施形態に係るイベント相関検出システム101は、到着率計算部102と、発生確率計算部103と、閾値計算部104と、イベント相関用ルールエンジン部160とを有する。
次に、上述した第1の実施形態を基本とする第2の実施形態について説明する。
(ただし、|E|は、イベントEを持つイベントが発生する回数、
λEは、上述したイベント発生頻度、
exp()は、自然対数を底とする指数関数を表す)。
(ただし、ΠE∈C_Pr(|E|>0|W)は、イベント型組み合わせCに属する任意のイベント型Eについて、Pr(|E|>0|W)を掛け合わせることを表す)。
(ただし、C(Nw,i)は、Nw個からi個を取り出す場合の組み合わせ数、
Nwは、イベント型ウィンドウ表における全ウィンドウ数、
W↑iは、べき乗を表す。即ち、W↑iは、Wをi回掛け合わせること、
iは、自然数を意味する)。
(ただし、(Σi>s_Nw×G(i))は、sよりも大きな値を持つiについて、Nw×G(i)の和をとること、
iは、自然数を表す)。
(ただし、|A|は、イベント型Aを持つイベントの発生回数を表す)。
(ただし、dR(Nw,W,R)(i)は、確率分布dR(Nw,W,R)のときの、値iを取る確率、
(Σi>t_Nw×dR(Nw,W,R)(i))は、tよりも大きな値を持つiについてNw×dR(Nw,W,R)(i)の和を算出すること、
iは、自然数を表す)。
(ただし、ΠE∈C_Pr(|E|>0|W)は、上記イベント型組み合わせCに属する全てのイベント型について、Pr(|E|>0|W)を掛け合わせること、
B(Nw,P)は、二項分布、即ち、
B(Nw,P)=Σi_C(Nw,i)×P↑i×(1-P)↑i、
(ただし、C(Nw,i)は、Nw個からi個を取り出す場合の組み合わせ数、
P↑iは、べき乗を表す。即ち、P↑iは、Wをi回掛け合わせることを意味する。
次に、発生確率計算部103は、イベント到着率表記憶部260を参照しながら、イベントごとに一定時間間隔W[sec]でのイベント発生確率を、下記のように計算し、その算出した値を発生確率表記憶部280に記憶する(ステップA220)。
「QueryError」,「NWCongestion」、
「QueryError」,「TooManyRequest」、
「DBError」,「NWCongestion」、
「DBError」,「TooManyRequest」、
「NWCongestion」,「TooManyRequest」。
「QueryError」,「TooManyRequest」,「DBError」。
TooManyRequest⇒QueryError。
図20は、本発明の各実施形態に係るイベント相関検出システムを実現可能な計算処理装置のハードウェア構成を、概略的に示すブロック図である。
(付記1)
複数のイベント型を対象として、
前記イベント型ごとに、前記イベント型が所定の時間間隔において到着する到着率を算出する到着率計算部と、
前記イベント型ごとに、前記到着率と前記所定の時間間隔とに基づいて、前記イベント型が発生する第1確率を算出する発生確率計算部と、
前記第1確率に基づいて、前記複数のイベント型が発生する第2確率を計算し、前記第2確率に基づいてイベント型組み合わせに対する最小支持度及び依存ルールに対する最小信頼度を算出する閾値計算部と、
前記最小支持度と前記最小信頼度に応じて前記複数のイベント型間における相関を検出するイベント相関用ルールエンジン部とを
備えるイベント相関検出システム。
前記イベント型を前記所定の時間間隔ごとに分類し、各前記所定の時間間隔と、分類した前記イベント型の集合を関連付けして、ウィンドウを生成するウィンドウ分割部をさらに備え、
前記到着率計算部は、前記ウィンドウに応じて計算を行う
付記1に記載のイベント相関検出システム。
前記ウィンドウに存在する複数の前記イベント型を組にし、その組を集めることによって、第1組み合わせを生成する組み合わせ生成部と、
前記第1組み合わせに含まれる前記組に存在するイベント型が全て出現する前記ウィンドウの個数と、全ての前記ウィンドウの個数との比である支持度を計算し、前記第1組み合わせから、前記支持度が前記最小支持度より大きな値を持つ特定の前記組を、第2組み合わせとして選抜する組み合わせ選抜部と、
前記第2組み合わせに存在する組におけるイベント型の中から、一つのイベント型を帰結とし、残りのイベント型を前提条件とする依存ルールを生成する依存ルール生成部と、
前記依存ルールに出現する前記イベント型が出現するウィンドウの個数と、前記前提条件が出現する前記ウィンドウの個数との比である信頼度を算出する依存ルール選抜部とを、
さらに備え、
前記イベント相関用ルールエンジン部は、前記信頼度と前記最小信頼度との大小に基づいて前記イベント型間における相関を検出する
付記2に記載のイベント相関検出システム。
前記組み合わせ生成部は、組み合わせ個数を2と設定し、
前記組み合わせ生成部は、前記組み合わせ個数分の前記イベント型を組み合わせて前記組を作成する第1処理を行い、
前記イベント相関用ルールエンジン部は、前記信頼度が前記最小信頼度よりも大きな値を持つ組が存在すると判定した場合には、前記第2組み合わせを第3組み合わせとして保持する第2処理を行ったのち、
前記組み合わせ個数を1増やす第3処理を行い、
前記イベント相関用ルールエンジン部が、前記信頼度が前記最小信頼度よりも大きな値を持つ組が存在しないと判定するまで、前記第1処理乃至前記第3処理を順次繰り返し、
前記イベント相関用ルールエンジン部は、前記信頼度が前記最小信頼度よりも大きな値を持つ組が存在しないと判定する場合には、前記イベント相関用ルールエンジン部が最後に保持した特定の前記第3組み合わせを出力する
付記3に記載のイベント相関検出システム。
前記閾値計算部は、
前記第1確率に基づいて、前記第2確率を計算し、その計算した値をまとめて確率分布G(i)を生成する確率分布計算部と、
前記確率分布G(i)(但しiは、自然数)においてiの値が特定値以上におけるG(i)の総和を取り、その算出した総和と全前記ウィンドウ数との積が、所定の想定誤検出率以下である範囲における最小の前記特定値を出力する分布確率計算部とを
有する付記2に記載のイベント相関検出システム。
前記閾値計算部は、
Pr(|E|>0|W)=1-exp(-λE×W)、
(ただし、Wは、前記ウィンドウにおける時間間隔、
λE=NE÷T、
exp()は、指数関数、
Tは、時間間隔、
NEは、時間間隔Tにおいて、イベント型Eが発生する回数、
「|E|」は、イベント型Eを持つイベントが発生する回数、
Pr(|E|>0|W)は、ウィンドウにおける時間間隔Wにおいて、イベント型Eを持つイベントを持つイベントが発生する確率である)、
の処理に従い、前記第1確率をあらわすPr(|E|>0|W)を計算し、前記第1確率に基づいて、前記第2確率を計算し、前記第2確率に基づいて前記最小支持度を算出する付記2に記載のイベント相関検出システム。
前記閾値計算部は、
Pr(R|W)=Pr(|A|>0|W)、
(ただし、Rは、依存ルール、
Aは、前記依存ルールにおける前記帰結、
Wは、前記ウィンドウの時間間隔、
「|A|」は、イベント型Aを持つイベントが発生する回数、
Pr(|A|>0|W)は、前記ウィンドウにおける時間間隔Wにおいて、イベント型Aを持つイベントを持つイベントが少なくとも1回以上発生する確率、
Pr(R|W)は、「前記ウィンドウにおける時間間隔Wにおいて、依存ルールRが発生する」確率である)、
の処理に従い前記第1確率を表すPr(R|W)を計算し、前記第1確率に基づいて、前記第2確率を計算し、前記第2確率に基づいて前記最小信頼度を算出する
付記2に記載のイベント相関検出システム。
前記閾値計算部は、前記ウィンドウに存在する前記時間間隔数と前記第1確率との2項分布の計算処理に従って、前記確率分布を計算し、前記確率分布に基づいて前記最小支持度を算出する
付記5に記載のイベント相関検出システム。
前記閾値計算部は、前記ウィンドウに存在する前記時間間隔数と前記第1確率との2項分布の計算処理に従って、前記確率分布を計算し、前記確率分布に基づいて前記最小信頼度を算出する付記5に記載のイベント相関検出システム。
前記閾値計算部は、
(Σi>s_Nw×G(i))<p0、
(ただし、Nwは、全前記ウィンドウの個数、
G(i)は、前記λEに基づいて算出した前記確率分布、
(Σi>s_Nw×G(i))は、tよりも大きなiについてNw×G(i)の総和値、
p0は、前記想定誤検出率、
iは、自然数である)、
を満たす最小のsを計算し、その算出値を前記最小支持度とする
付記5に記載のイベント相関検出システム。
前記閾値計算部は、
(Σi>t_Nw×dR(Nw,W,R)(i))<p0
(ただし、Nwは、全前記ウィンドウの個数、
dR(Nw,W,R)(i)は、前記依存ルールRに基づいて算出した前記確率分布、
(Σi>t_Nw×dR(Nw,W,R)(i))は、tよりも大きなiについてNw×dR(Nw,W,R)(i)の総和値、
p0は、前記想定誤検出率、
iは、自然数である)、
を満たす最小のtを計算し、その算出値を前記最小信頼度とする
付記5に記載のイベント相関検出システム。
複数のイベント型を対象として、
前記イベント型ごとに、前記イベント型が所定の時間間隔において到着する到着率を計算し、
前記イベント型ごとに、前記到着率と前記時間間隔とに基づいて、前記イベント型が発生する第1確率を計算し、
前記第1確率に基づいて、前記複数のイベント型が発生する第2確率を計算し、前記第2確率に基づいて、イベント型組み合わせに対する最小支持度及び依存ルールに対する最小信頼度を計算し、
前記最小支持度と前記最小信頼度に応じて前記複数のイベント型間における相関を検出する
イベント相関検出方法。
複数のイベント型を対象として、
前記イベント型ごとに、前記イベント型が所定の時間間隔において到着する到着率を算出する到着率機能と、
前記イベント型ごとに、前記到着率と前記時間間隔とに基づいて、前記イベント型が発生する第1確率を算出する発生確率計算機能と、
前記第1確率に基づいて、前記複数のイベント型が発生する第2確率を計算し、前記第2確率に基づいて、イベント型組み合わせに対する最小支持度及び依存ルールに対する最小信頼度を算出する閾値計算機能と、
前記最小支持度と前記最小信頼度に応じて前記複数のイベント型間における相関を検出するイベント相関用ルールエンジン機能とを
コンピュータに実現させるコンピュータ・プログラム。
26 マイニングカーネルインタフェース機能
28 マイニング結果記録機能
101 イベント相関検出システム
102 到着率計算部
103 発生確率計算部
104 閾値計算部
105 イベント相関検出システム
110 イベント発生源
115 GUI
120 想定誤検出率記憶部
130 イベント履歴DB記憶部
140 ウィンドウ分割部
150 イベント型ウィンドウ表記憶部
160 イベント相関用ルールエンジン部
170 イベント相関規則記憶部
210 イベント型ウィンドウ分割部
220 イベント型ウィンドウ表記憶部
230 規則生成部
240 ルール候補記憶部
260 イベント到着率表記憶部
280 発生確率表記憶部
310 閾値記憶部
410 組み合わせ生成部
420 イベント組み合わせ記憶部
430 最小支持度計算部
440 最小支持度記憶部
450 組み合わせ選抜部
460 依存ルール生成部
470 依存ルール記憶部
480 最小信頼度計算部
490 最小信頼度記憶部
510 依存ルール選抜部
610 確率分布計算部
620 確率分布記憶部
630 分布確率逆計算部
710 確率分布計算部
720 確率分布記憶部
730 分布確率逆計算部
2301 CPU
2302 メモリ
2303 ディスク
2304 出力装置
2305 入力装置
2306 計算処理装置
2307 不揮発性記録媒体
1001 ウィンドウ分割部
1002 イベント型ウィンドウ分割部
1003 イベント相関用ルールエンジン部
1004 イベント履歴Database記憶部
1005 イベント型ウィンドウ分割部
1006 イベント型ウィンドウ表記憶部
930 規則生成部
1008 イベント相関規則記憶部
Claims (10)
- 複数のイベント型を対象として、
前記イベント型ごとに、前記イベント型が所定の時間間隔において到着する到着率を計算する到着率計算部と、
前記イベント型ごとに、前記到着率と前記時間間隔とに基づいて、前記イベント型が発生する第1確率を計算する発生確率計算部と、
前記第1確率に基づいて、前記複数のイベント型が発生する第2確率を計算し、前記第2確率に基づいて、イベント型組み合わせに対する最小支持度及び依存ルールに対する最小信頼度を計算する閾値計算部と、
前記最小支持度と前記最小信頼度に応じて前記複数のイベント型間における相関を検出するイベント相関用ルールエンジン部とを
備えることを特徴とするイベント相関検出システム。 - 前記イベント型を前記時間間隔ごとに分類し、各前記所定の時間間隔と、分類した前記イベント型の集合を関連付けして、ウィンドウを生成するウィンドウ分割部をさらに備え、
前記到着率計算部は、前記ウィンドウに応じて計算を行う
ことを特徴とする請求項1に記載のイベント相関検出システム。 - 前記ウィンドウに存在する複数の前記イベント型を組にし、その組を集めることによって、第1組み合わせを生成する組み合わせ生成部と、
前記第1組み合わせに含まれる前記組に存在するイベント型が全て出現する前記ウィンドウの個数と、全ての前記ウィンドウの個数との比である支持度を計算し、前記第1組み合わせから、前記支持度が前記最小支持度より大きな値を持つ特定の前記組を、第2組み合わせとして選抜する組み合わせ選抜部と、
前記第2組み合わせに存在する組におけるイベント型の中から、一つのイベント型を帰結とし、残りのイベント型を前提条件とする依存ルールを生成する依存ルール生成部と、
前記依存ルールに出現する前記イベント型が出現するウィンドウの個数と、前記前提条件が出現する前記ウィンドウの個数との比である信頼度を計算する依存ルール選抜部とを、
さらに備え、
前記イベント相関用ルールエンジン部は、前記信頼度と前記最小信頼度との大小に基づいて前記イベント型間における相関を検出する
ことを特徴とする請求項2に記載のイベント相関検出システム。 - 前記組み合わせ生成部は、組み合わせ個数を2と設定し、
前記組み合わせ生成部は、前記組み合わせ個数分の前記イベント型を組み合わせて前記組を作成する第1処理を行い、
前記イベント相関用ルールエンジン部は、前記信頼度が前記最小信頼度よりも大きな値を持つ組が存在すると判定した場合には、前記第2組み合わせを第3組み合わせとして保持する第2処理を行ったのち、
前記組み合わせ個数を1増やす第3処理を行い、
前記イベント相関用ルールエンジン部が、前記信頼度が前記最小信頼度よりも大きな値を持つ組が存在しないと判定するまで、前記第1処理乃至前記第3処理を順次繰り返し、
前記イベント相関用ルールエンジン部は、前記信頼度が前記最小信頼度よりも大きな値を持つ組が存在しないと判定した場合には、前記イベント相関用ルールエンジン部が最後に保持した特定の前記第3組み合わせを出力する
ことを特徴とする請求項3に記載のイベント相関検出システム。 - 前記閾値計算部は、
前記第1確率に基づいて、前記第2確率を計算し、その計算した値をまとめて確率分布G(i)(但しiは、自然数)を生成する確率分布計算部と、
前記確率分布G(i)においてiの値が特定値以上におけるG(i)の総和を取り、その算出した総和と全前記ウィンドウ数との積が、所定の想定誤検出率以下である範囲における最小の前記特定値を出力する分布確率計算部と、
から構成されることを特徴とする請求項2乃至4のいずれか1項に記載のイベント相関検出システム。 - 前記閾値計算部は、
Pr(|E|>0|W)=1-exp(-λE×W)、
(ただし、Wは、前記ウィンドウにおける時間間隔、
λE=NE÷T、
exp()は、指数関数、
Tは、時間間隔、
NEは、時間間隔Tにおいて、イベント型Eが発生する回数、
「|E|」は、イベント型Eを持つイベントが発生する回数、
Pr(|E|>0|W)は、ウィンドウにおける時間間隔Wにおいて、イベント型Eを持つイベントを持つイベントが発生する確率である)、
の処理に従い、前記第1確率をあらわすPr(|E|>0|W)を計算し、前記第1確率に基づいて、前記第2確率を計算し、前記第2確率に基づいて前記最小支持度を計算する
ことを特徴とする、請求項2乃至5のいずれか1項に記載のイベント相関検出システム。 - 前記閾値計算部は、
Pr(R|W)=Pr(|A|>0|W)、
(ただし、Rは、依存ルール、
Aは、前記依存ルールにおける前記帰結、
Wは、前記ウィンドウの時間間隔、
「|A|」は、イベント型Aを持つイベントが発生する回数、
Pr(|A|>0|W)は、前記ウィンドウにおける時間間隔Wにおいて、イベント型Aを持つイベントを持つイベントが少なくとも1回以上発生する確率、
Pr(R|W)は、「前記ウィンドウにおける時間間隔Wにおいて、依存ルールRが発生する」確率である)、
の処理に従い前記第1確率を表すPr(R|W)を計算し、前記第1確率に基づいて、前記第2確率を計算し、前記第2確率に基づいて前記最小信頼度を計算する
ことを特徴とする、請求項2乃至6のいずれか1項に記載のイベント相関検出システム。 - 前記閾値計算部は、
(Σi>s_Nw×G(i))<p0、
(ただし、Nwは、全前記ウィンドウの個数、
G(i)は、前記λEに基づいて算出した前記確率分布、
(Σi>s_Nw×G(i))は、tよりも大きなiについてNw×G(i)の総和値、
p0は、前記想定誤検出率、
iは、自然数である)、
を満たす最小のsを計算し、その算出値を前記最小支持度とする
ことを特徴とする請求項5あるいは6のいずれか1項に記載のイベント相関検出システム。 - 複数のイベント型を対象として、
前記イベント型ごとに、前記イベント型が所定の時間間隔において到着する到着率を計算し、
前記イベント型ごとに、前記到着率と前記時間間隔とに基づいて、前記イベント型が発生する第1確率を計算し、
前記第1確率に基づいて、前記複数のイベント型が同時に発生する第2確率を計算し、前記第2確率に基づいて、イベント型組み合わせに対する最小支持度及び依存ルールに対する最小信頼度を計算し、
前記最小支持度と前記最小信頼度に応じて前記複数のイベント型間における相関を検出する
ことを特徴とするイベント相関検出方法。 - 複数のイベント型を対象として、
前記イベント型ごとに、前記イベント型が所定の時間間隔において到着する到着率を計算する到着率機能と、
前記イベント型ごとに、前記到着率と前記時間間隔とに基づいて、前記イベント型が発生する第1確率を計算する発生確率計算機能と、
前記第1確率に基づいて、前記複数のイベント型が同時に発生する第2確率を計算し、前記第2確率に基づいて、イベント型組み合わせに対する最小支持度及び依存ルールに対する最小信頼度を計算する閾値計算機能と、
前記最小支持度と前記最小信頼度に応じて前記複数のイベント型間における相関を検出するイベント相関用ルールエンジン機能とを
コンピュータに実現させるコンピュータ・プログラム。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US14/388,589 US20150058272A1 (en) | 2012-03-26 | 2013-03-08 | Event correlation detection system |
JP2014507378A JP6060969B2 (ja) | 2012-03-26 | 2013-03-08 | イベント相関検出システム |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2012-068639 | 2012-03-26 | ||
JP2012068639 | 2012-03-26 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2013145584A1 true WO2013145584A1 (ja) | 2013-10-03 |
Family
ID=49258902
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2013/001481 WO2013145584A1 (ja) | 2012-03-26 | 2013-03-08 | イベント相関検出システム |
Country Status (3)
Country | Link |
---|---|
US (1) | US20150058272A1 (ja) |
JP (1) | JP6060969B2 (ja) |
WO (1) | WO2013145584A1 (ja) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104699019A (zh) * | 2013-12-09 | 2015-06-10 | 中芯国际集成电路制造(上海)有限公司 | 机台恢复检验系统以及机台恢复检验方法 |
JP2018028778A (ja) * | 2016-08-17 | 2018-02-22 | 日本電信電話株式会社 | パターン抽出及びルール生成装置、及びその方法 |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US10462199B2 (en) * | 2016-12-23 | 2019-10-29 | Cerner Innovation, Inc. | Intelligent and near real-time monitoring in a streaming environment |
JP2020071570A (ja) * | 2018-10-30 | 2020-05-07 | ファナック株式会社 | データ作成装置、デバッグ装置、データ作成方法及びデータ作成プログラム |
CN110932272A (zh) * | 2019-12-13 | 2020-03-27 | 国网福建省电力有限公司三明供电公司 | 一种三遥配电终端优化配置方法 |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH10134086A (ja) * | 1996-08-30 | 1998-05-22 | Kokusai Denshin Denwa Co Ltd <Kdd> | 因果関係検出装置及び方法 |
JP2006004346A (ja) * | 2004-06-21 | 2006-01-05 | Fujitsu Ltd | パターン検出プログラム |
JP2008299690A (ja) * | 2007-06-01 | 2008-12-11 | Honda Motor Co Ltd | イベント解析装置 |
JP2009217567A (ja) * | 2008-03-11 | 2009-09-24 | Oki Electric Ind Co Ltd | ログデータ相関分析装置及び方法 |
WO2010131746A1 (ja) * | 2009-05-15 | 2010-11-18 | 日本電気株式会社 | 障害原因推定システム、障害原因推定方法、及び障害原因推定プログラム |
JP2011159125A (ja) * | 2010-02-01 | 2011-08-18 | Nec Corp | イベントクラスタリングシステム、そのコンピュータプログラムおよびデータ処理方法 |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120078903A1 (en) * | 2010-09-23 | 2012-03-29 | Stefan Bergstein | Identifying correlated operation management events |
US20130027561A1 (en) * | 2011-07-29 | 2013-01-31 | Panasonic Corporation | System and method for improving site operations by detecting abnormalities |
-
2013
- 2013-03-08 JP JP2014507378A patent/JP6060969B2/ja active Active
- 2013-03-08 WO PCT/JP2013/001481 patent/WO2013145584A1/ja active Application Filing
- 2013-03-08 US US14/388,589 patent/US20150058272A1/en not_active Abandoned
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH10134086A (ja) * | 1996-08-30 | 1998-05-22 | Kokusai Denshin Denwa Co Ltd <Kdd> | 因果関係検出装置及び方法 |
JP2006004346A (ja) * | 2004-06-21 | 2006-01-05 | Fujitsu Ltd | パターン検出プログラム |
JP2008299690A (ja) * | 2007-06-01 | 2008-12-11 | Honda Motor Co Ltd | イベント解析装置 |
JP2009217567A (ja) * | 2008-03-11 | 2009-09-24 | Oki Electric Ind Co Ltd | ログデータ相関分析装置及び方法 |
WO2010131746A1 (ja) * | 2009-05-15 | 2010-11-18 | 日本電気株式会社 | 障害原因推定システム、障害原因推定方法、及び障害原因推定プログラム |
JP2011159125A (ja) * | 2010-02-01 | 2011-08-18 | Nec Corp | イベントクラスタリングシステム、そのコンピュータプログラムおよびデータ処理方法 |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104699019A (zh) * | 2013-12-09 | 2015-06-10 | 中芯国际集成电路制造(上海)有限公司 | 机台恢复检验系统以及机台恢复检验方法 |
JP2018028778A (ja) * | 2016-08-17 | 2018-02-22 | 日本電信電話株式会社 | パターン抽出及びルール生成装置、及びその方法 |
Also Published As
Publication number | Publication date |
---|---|
US20150058272A1 (en) | 2015-02-26 |
JP6060969B2 (ja) | 2017-01-18 |
JPWO2013145584A1 (ja) | 2015-12-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10902062B1 (en) | Artificial intelligence system providing dimension-level anomaly score attributions for streaming data | |
JP6060969B2 (ja) | イベント相関検出システム | |
JP5532150B2 (ja) | 運用管理装置、運用管理方法、及びプログラム | |
JP2007096796A (ja) | ネットワーク障害診断装置、ネットワーク障害診断方法およびネットワーク障害診断プログラム | |
US20130191107A1 (en) | Monitoring data analyzing apparatus, monitoring data analyzing method, and monitoring data analyzing program | |
JP6665784B2 (ja) | ログ分析システム、ログ分析方法およびログ分析プログラム | |
JP6521096B2 (ja) | 表示方法、表示装置、および、プログラム | |
JP2005302028A (ja) | プログラムを計装するプローブ最適化のための方法およびシステム | |
JP5051135B2 (ja) | 資源情報収集装置、資源情報収集方法、プログラム、および、収集スケジュール生成装置 | |
US9658834B2 (en) | Program visualization device, program visualization method, and program visualization program | |
US8660979B2 (en) | Event prediction | |
US8924794B2 (en) | Method and computer program product for forecasting system behavior | |
US20090187533A1 (en) | Automatically identifying an optimal set of attributes to facilitate generating best practices for configuring a networked system | |
JP2000194745A (ja) | トレンド評価装置及びトレンド評価方法 | |
JP5439775B2 (ja) | 障害対応プログラム、障害対応装置、及び障害対応システム | |
JP2021174473A (ja) | ユーザに提案する材料を決定するシステム | |
JP2016085152A (ja) | 診断装置、診断プログラム及び診断方法 | |
JP2018014000A (ja) | テスト支援プログラム、テスト支援装置、及びテスト支援方法 | |
US11113268B2 (en) | Method and device for restoring missing operational data | |
CN113434326A (zh) | 基于分布式集群拓扑实现网络系统故障定位的方法及装置、处理器及其计算机可读存储介质 | |
JP2022115316A (ja) | ログ検索支援装置、及びログ検索支援方法 | |
JP6396615B1 (ja) | 情報処理プログラム、情報処理装置及びデバッグシステム | |
JP2017207878A (ja) | 欠落データ推定方法、欠落データ推定装置および欠落データ推定プログラム | |
JP2000242651A (ja) | データマイニング方法およびデータマイニング装置 | |
KR20190123369A (ko) | 머신러닝 기반 악성코드 탐지를 위한 특성선정 방법 및 이를 수행하기 위한 기록매체 및 장치 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 13768221 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2014507378 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 14388589 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 13768221 Country of ref document: EP Kind code of ref document: A1 |