Border event prediction method based on affair graph

Abstract

The invention discloses a border event prediction method based on a affair atlas, which comprises the following steps: establishing the affair atlas according to a historical event set family, and counting the prior probability of occurrence of a subsequent event type corresponding to each event type in the affair atlas; based on the prior probability of occurrence of the subsequent event type corresponding to each event type, performing pruning operation on the relationship in the affair graph, and obtaining the new prior probability of occurrence of the subsequent event type corresponding to each event type in the affair graph after pruning; based on the adjoint large event corresponding to each event, optimizing the event graph; and on the basis of the current occurrence event, according to the new prior probability of occurrence of the subsequent event type corresponding to each event type in the optimized event graph, predicting the type of the subsequent event which is about to occur. According to the method, the border event prediction accuracy is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of event prediction, and more specifically to a method for predicting border events based on an event graph. Background Art

[0002] In the border area, a series of border incidents usually occur, and these incidents often have a certain sequence and causal relationship. If a model can be established for the prediction and early warning of border incidents, and the model can be used to analyze the sequence and causal relationship between border incidents and other events related to the incidents, it will be of guiding significance for the next actions of the border area forces.

[0003] As a method to describe the evolutionary laws and patterns between events, the event graph is usually used in risk management and event situation prediction, which is consistent with the border event prediction scenario. However, the current research on using this technology to predict border events is relatively niche, and all of them use text mining technology to discover the connection rules between events, with limited prediction accuracy. As for the border event set, there are few related research reports on using statistical analysis methods to mine the internal connections between events and using graphs for prediction.

[0004] Therefore, how to improve the accuracy of border event prediction based on the event graph is an urgent problem that technical personnel in this field need to solve. Summary of the invention

[0005] In view of the above problems, the present invention provides a border event prediction method based on a matter graph to at least solve some of the technical problems mentioned in the above background technology.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] The present invention provides a border event prediction method based on a fact map, comprising the following steps:

[0008] S1. Establish a graph of events based on historical event clusters, and calculate the prior probability of the occurrence of subsequent event types corresponding to each event type in the graph of events;

[0009] S2. Based on the prior probability of the occurrence of the subsequent event type corresponding to each event type, prune the relationship in the event graph, and obtain the new prior probability of the occurrence of the subsequent event type corresponding to each event type in the event graph after pruning;

[0010] S3. Optimize the event graph based on the accompanying major events corresponding to each event;

[0011] S4. Based on the current event, the new prior probability of the subsequent event type corresponding to each event type in the optimized event graph is used to predict the subsequent event type that is about to occur.

[0012] Furthermore, the step S1 specifically includes:

[0013] S11. Let there be a total of c event sets in the historical event set I, expressed as ; Select the i-th event set from the historical event family I , i=1; create a blank event graph;

[0014] S12. From the event set Select the jth event , and for events Make a judgment:

[0015] If the event does not exist in the event graph The corresponding type, then the event After adding the corresponding type to the event graph, check the event again Subsequent events ;

[0016] If there is an event in the event graph The corresponding type, then view the event Subsequent events ;

[0017] S13. Events Make a judgment:

[0018] If the event does not exist in the event graph The corresponding type, then the event The corresponding types are added to the event graph, and the events in the event graph are Types and events The type of relationship is established, and the corresponding contact value ;

[0019] If there is an event in the event graph The corresponding type is directly related to the event in the event graph. Types and events The type of relationship is established, and the corresponding contact value ;

[0020] S14, loop steps S12 to S13 until the current event set is traversed ;

[0021] S15, let i=i+1, and repeat steps S12-S14 until all historical event sets I are traversed;

[0022] S16. Take each event type in the event graph as a node, and the subsequent event type corresponding to each event type as a successor node; traverse each node a in the event graph, and check all successor nodes of node a; record all successor nodes of node a as successor node set b, and let the successor node set b have a total of n successor nodes, expressed as ;

[0023] Get the total value of the connection between node a and the set of subsequent nodes b, expressed as:

[0024]

[0025] Among them, L a Represents the total value of the connection between node a and the set of subsequent nodes b; b f represents the fth successor node in the successor node set b; Represents node a and subsequent node b f The connection value between them;

[0026] Get the prior probability of node a and each successor node in the successor node set b, expressed as:

[0027]

[0028] in, For node a and the subsequent node b f The prior probability is used to indicate that after node a occurs, the subsequent node b occurs f The prior probability of .

[0029] Furthermore, the step S2 specifically includes:

[0030] S21. Let the event graph contain m nodes, represented as ; If there is a node in the event graph N Its follow-up node The prior probability between is less than the first threshold, then the node is obtained and rear drive nodes The posterior probability is expressed as:

[0031]

[0032] in, For Node and rear drive nodes The posterior probability is used to indicate that after the subsequent node After that, the node The posterior probability of Representation Node and rear drive nodes The contact value of Represents the zth node in the event graph N; Representation Node and nodes The contact value of

[0033] S22, if the node and rear drive nodes If the posterior probability of is less than the second threshold, then the node and rear drive nodes Prune the relationship between

[0034] S23, repeating steps S21-S22 until all the relationship values ​​corresponding to the prior probabilities less than the first threshold in the event graph N are pruned;

[0035] S24. Recalculate the prior probability of each node and its successor node, record it as the new prior probability, expressed as:

[0036]

[0037] in, For Node and rear drive nodes The new prior probability is used to indicate that when a node After that, the rear drive node occurs The prior probability of Representation Node and driver node The contact value of .

[0038] Furthermore, the step S3 specifically includes:

[0039] S31. An event composed of multiple events is called a major event. The set of major events is denoted as S. Let there be g major events in total, expressed as ;

[0040] The major event that occurs after each event is called the accompanying major event, and the accompanying major event cluster corresponding to each node in the event graph N is denoted as NF, and ,in, Represents the set of accompanying large events corresponding to the mth node in the event graph N;

[0041] The set containing all events is called the complete set E, and there are k events in it, which can be expressed as ; Among them, the currently selected event is , h=1;

[0042] S32. Count the selected events in the current round All the accompanying major events of ; Screen the selected events from the event map N The corresponding node , and the selection event All accompanying major events are added to the node A collection of accompanying major events middle;

[0043] S33. Statistics accompanying major events collection The number of occurrences of each major event type in the event tree. If the number of occurrences of a major event type is greater than the preset value, the major event type is used as a node. Frequent itemsets of ;

[0044] S34. Let h=h+1, repeat the above steps S32~S33, obtain the frequent item set corresponding to each node in the event graph N; and add the frequent item set information to the event graph N.

[0045] Further, in step S32, the statistics of selected events in the current round The accompanying major events include:

[0046] Traverse the set of major events S and select the major events that meet the preset conditions as selected events The pre-conditions include:

[0047] (1) Regions where major events occur and selected events The occurrence area is consistent;

[0048] (2) Select an event Occurs within the time interval of a major event; or, the time interval between the end of a major event and the selected event The duration is less than the preset duration;

[0049] (3) Major events do not include selected events .

[0050] Furthermore, the step S4 specifically includes:

[0051] S41, record the current event as event e, traverse the event graph, and obtain:

[0052] (1) Node corresponding to event e ;

[0053] (2) Node The set of rear-drive nodes , let the set of rear-drive nodes There are s rear-drive nodes in ; According to the new prior probability, the set of follower nodes Sort each successor node in

[0054] (3)Set S of major event types occurring within a preset time period e ;

[0055] S42. Select a successor node set and choose a successor node , where l = 2;

[0056] S43. Select the q-th successor node in the successor node set , where q = 1; S44. Compare the successor node

[0057] with the successor node . If l > s, then execute step S45; If

[0058] , then make the following judgment: If

[0059] (1)If the successor node has a frequent item set belonging to the set S of major event types e , then make the following judgment:

[0060] If the successor node does not have a frequent item set belonging to the set S of major event types e , then insert the successor node into the q-th position; let l = l + 1, and loop to execute steps S43 - S44;

[0061] If the successor node has a frequent item set belonging to the set S of major event types e , then judge the new prior probability of event e and the successor node . If the new prior probability of event e and the successor node is greater than the new prior probability of event e and the successor node , then insert the successor node into the q-th position; let l = l + 1, and loop to execute steps S43 - S44; if the new prior probability of event e and the successor node is less than or equal to the new prior probability of event e and the successor node , then let q = q + 1. If q = l, then let l = l + 1, and loop to execute steps S43 - S44; if q < l, loop to execute step S44;

[0062] (2)If the successor node does not have a frequent item set belonging to the set S of major event types e , then make the following judgment:

[0063] If the successor node has frequent itemsets belonging to the large event type set S e , then let q = q + 1. If q = l, then let l = l + 1, and loop to execute steps S43 - S44; if q < l, loop to execute step S44;

[0064] If the successor node has no frequent itemsets belonging to the large event type set S e , then judge the new prior probabilities of event e and the successor node . If the new prior probabilities of event e and the successor node are greater than the new prior probabilities of event e and the successor node , then insert the successor node into the q - th position; let l = l + 1, and loop to execute steps S43 - S44; if the new prior probabilities of event e and the successor node are less than or equal to the new prior probabilities of event e and the successor node , then let q = q + 1. If q = l, then let l = l + 1, and loop to execute steps S43 - S44; if q < l, loop to execute step S44;

[0065] S45. End the loop. At this time, all successor nodes in the successor node set are sorted according to the predicted occurrence probability.

[0066] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses a border event prediction method based on an event logic graph, having the following beneficial effects:

[0067] The present invention proposes a method for establishing an event logic graph based on an existing historical event set family; prunes the event logic graph based on the probability distribution of precursor events; comprehensively considers the large event set accompanied by events, and uses the event logic graph for event prediction, effectively improving the accuracy of border event prediction.

[0068] The technical solutions of the present invention will be further described in detail below through the accompanying drawings and embodiments. Brief Description of the Drawings

[0069] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, other accompanying drawings can be obtained according to the provided accompanying drawings without creative efforts.

[0070] Figure 1A schematic flow chart of a method for predicting border events based on a causal graph provided in an embodiment of the present invention.

[0071] Figure 2 A schematic diagram of the process of constructing a logical graph provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0072] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0073] The embodiment of the present invention discloses a border event prediction method based on a matter-of-fact graph, see Figure 1 As shown, the following steps are included:

[0074] S1. Establish a graph of events based on historical event clusters, and calculate the prior probability of the occurrence of subsequent event types corresponding to each event type in the graph of events;

[0075] S2. Based on the prior probability of the occurrence of the subsequent event type corresponding to each event type, prune the relationship in the event graph, and obtain the new prior probability of the occurrence of the subsequent event type corresponding to each event type in the event graph after pruning;

[0076] S3. Optimize the event graph based on the accompanying major events corresponding to each event;

[0077] S4. Based on the current event, the new prior probability of the subsequent event type corresponding to each event type in the optimized event graph is used to predict the subsequent event type that is about to occur.

[0078] Next, each of the above steps will be described in detail.

[0079] In the above step S1, a fact map is established based on the historical event cluster, and the prior probability of the occurrence of the subsequent event type corresponding to each event type in the fact map is counted;

[0080] Specifically, in border control, various events may occur, and there may be certain sequence and causal relationships between events. Usually, these related events are grouped together to represent a certain type of specific theme. Many events are aggregated according to their themes to form an event set, which can also be called a large event. By statistically analyzing the sequence of events, it is possible to predict the next event when a certain type of event occurs. In the embodiment of the present invention, a probability statistical method is used to count the probabilities of other events occurring after each event, and this relationship is represented in the form of a causal map. For specific methods, see. Figure 2 As shown, specifically including:

[0081] S11. Let there be a total of c event sets in the historical event set I, expressed as ; Select the i-th event set from the historical event family I , i=1; create a blank event graph;

[0082] S12. From the event set Select the jth event , and for events Make a judgment:

[0083] If the event does not exist in the event graph The corresponding type, then the event After adding the corresponding type to the event graph, check the event again Subsequent events ;

[0084] If there is an event in the event graph The corresponding type, then view the event Subsequent events ;

[0085] S13. Events Make a judgment:

[0086] If the event does not exist in the event graph The corresponding type, then the event The corresponding types are added to the event graph, and the events in the event graph are Types and events The type of relationship is established, and the corresponding contact value , indicating the correlation between these two types of events;

[0087] If there is an event in the event graph The corresponding type is directly related to the event in the event graph. Types and events The type of relationship is established, and the corresponding contact value ;

[0088] S14, loop steps S12 to S13 until the current event set is traversed ;

[0089] S15, let i=i+1, and repeat steps S12-S14 until all historical event sets I are traversed;

[0090] S16. Take each event type in the event graph as a node, and the subsequent event type corresponding to each event type as a successor node; traverse each node a in the event graph, and check all successor nodes of node a; record all successor nodes of node a as successor node set b, and let the successor node set b have a total of n successor nodes, expressed as ;

[0091] Get the total value of the connection between node a and the set of subsequent nodes b, expressed as:

[0092]

[0093] Among them, L a Represents the total value of the connection between node a and the set of subsequent nodes b; b f represents the fth successor node in the successor node set b; Represents node a and subsequent node b f The connection value between them;

[0094] Get the prior probability of node a and each successor node in the successor node set b, expressed as:

[0095]

[0096] in, For node a and the subsequent node b f The prior probability is used to indicate that after node a occurs, the subsequent node b occurs f The prior probability of .

[0097] In the above step S2, based on the prior probability of occurrence of the subsequent event type corresponding to each event type, the relationship in the event graph is pruned, and the new prior probability of occurrence of the subsequent event type corresponding to each event type in the event graph after pruning is obtained;

[0098] Specifically, under normal circumstances, there is a certain correlation between the order in which border events occur. When a border event occurs, it is strongly correlated with a previous event, but this event is not necessarily a neighboring preceding event. To calculate the events that may occur after an event, it is necessary not only to consider the distribution of subsequent events of each event, but also to consider whether the subsequent event is related to the event. Here, by calculating the posterior probability of the subsequent event, the event graph obtained in the first step is pruned. While reducing the complexity of the graph, the correlation between the subsequent event and the preceding event is analyzed to improve the accuracy of the probability prediction. The specific steps are as follows:

[0099] S21. Let the event graph contain m nodes, represented as ; Obtain the connection value between each node in the event graph N and each of its subsequent nodes, which can be obtained through the above step S1;

[0100] If there is a node in the event graph N Its follow-up node The prior probability between is less than the first threshold (the first threshold <0.3), then the node is obtained and rear drive nodes The posterior probability is expressed as:

[0101]

[0102] in, For Node and rear drive nodes The posterior probability is used to indicate that after the subsequent node After that, the node The posterior probability of Representation Node and rear drive nodes The contact value of Represents the zth node in the event graph N; Representation Node and nodes The contact value of

[0103] S22, if the node and rear drive nodes The posterior probability is less than the second threshold ,Right now , then the node and rear drive nodes There is no correlation at all. At this time, the node in the event graph and rear drive nodes Prune the relationship between

[0104] S23, repeating steps S21-S22 until all the relationship values ​​corresponding to the prior probabilities less than the first threshold in the event graph N are pruned;

[0105] S24. Recalculate the prior probability of each node and its successor node, record it as the new prior probability, expressed as:

[0106]

[0107] in, For Node and rear drive nodes The new prior probability is used to indicate that when a node After that, the rear drive node occurs The prior probability of Representation Node and driver node The contact value of .

[0108] In the above step S3, the event graph is optimized based on the accompanying major events corresponding to each event;

[0109] Specifically, in border events, a small event is often accompanied by other large events, because some large events are one of the causes of small events. Therefore, in the event graph, in addition to counting the probability relationship between general events, we also need to use the co-occurrence of large events to supplement the prediction method. The core idea is to count the frequent item sets of each small event that accompanies the large event. Specifically, it includes:

[0110] S31. An event composed of multiple events is called a major event. The set of major events is denoted as S. Let there be g major events in total, expressed as ;

[0111] The major event that occurs after each event is called the accompanying major event, and the accompanying major event cluster corresponding to each node in the event graph N is denoted as NF, and ,in, Represents the set of accompanying large events corresponding to the mth node in the event graph N;

[0112] The set of major event types is denoted as ST, and it contains a total of x major event types, expressed as ;

[0113] The set containing all events is called the complete set E, and there are k events in it, which can be expressed as ; Among them, the currently selected event is , h=1;

[0114] S32. Count the selected events in the current round All the accompanying major events of ; select the selected events from the event graph N The corresponding node , and select the event All accompanying major events are added to the node A collection of accompanying major events middle;

[0115] Among them, statistics on selected events in the current round The accompanying major event specifically includes: traversing the major event set S, and taking the major events that meet the preset conditions as the selected events The preconditions include:

[0116] (1) Regions where major events occur and selected events The occurrence area is consistent;

[0117] (2) Select an event Occurs within the time interval of a major event; or, the time interval between the end of a major event and the selected event The duration is less than the preset duration;

[0118] (3) Major events do not include selected events .

[0119] S33. Statistics accompanying major events collection The number of occurrences of each major event type in the event tree. If the number of occurrences of a major event type is greater than the preset value, the major event type is used as a node. Frequent itemsets of ;

[0120] S34, let h=h+1, repeat the above steps S32~S33, obtain the frequent item set corresponding to each node in the event graph N; and add the frequent item set information to the event graph N as the basis for event prediction.

[0121] In the above step S4, based on the current event, according to the new prior probability of the subsequent event type corresponding to each event type in the optimized event graph, the subsequent event type to be occurred is predicted; specifically, it includes:

[0122] S41, record the current event as event e, traverse the event graph, and obtain:

[0123] (1) Node corresponding to event e ;

[0124] (2) Node The set of rear-drive nodes , let the set of rear-drive nodes There are s rear-drive nodes in ; Sort each successor node in the successor node set according to the new prior probability in the set;

[0125] (3)Set S of major event types that occurred within a preset recent time period e ;

[0126] S42. Select a successor node in the successor node set , l = 2;

[0127] S43. Select the q-th successor node in the set; , q = 1;

[0128] S44. Compare the successor node with the successor node . If l > s, execute step S45;

[0129] If , then make the following judgment:

[0130] (1)If the successor node has a frequent item set belonging to the set S of major event types e , then make the following judgment:

[0131] If the successor node does not have a frequent item set belonging to the set S of major event types e , then insert the successor node into the q-th position; Let l = l + 1, and loop to execute steps S43 - S44;

[0132] If the successor node has a frequent item set belonging to the set S of major event types e , then judge the new prior probability of event e and the successor node . If the new prior probability of event e and the successor node is greater than the new prior probability of event e and the successor node , then insert the successor node into the q-th position; Let l = l + 1, and loop to execute steps S43 - S44; If the new prior probability of event e and the successor node is less than or equal to the new prior probability of event e and the successor node , then let q = q + 1. If q = l, then let l = l + 1, and loop to execute steps S43 - S44; If q < l, loop to execute step S44;

[0133] (2)If the successor node does not have a frequent item set belonging to the set S of major event types e, the following judgments are made:

[0134] If the successor node has a frequent item set belonging to the large event type set S e , then let q = q + 1. If q = l, then let l = l + 1, and loop to execute steps S43~S44; if q < l, loop to execute step S44;

[0135] If the successor node does not have a frequent item set belonging to the large event type set S e , then judge the new prior probability of event e and the successor node ; if the new prior probability of event e and the successor node is greater than the new prior probability of event e and the successor node , then insert the successor node into the qth position; let l = l + 1, and loop to execute steps S43~S44; if the new prior probability of event e and the successor node is less than or equal to the new prior probability of event e and the successor node , then let q = q + 1. If q = l, then let l = l + 1, and loop to execute steps S43~S44; if q < l, loop to execute step S44;

[0136] S45. End the loop. At this time, all the successor nodes in the successor node set are sorted according to the predicted occurrence probability.

[0137] In summary, the embodiment of the present invention provides a border event prediction method based on an event logic graph. For the existing event data set, a method for establishing an event logic graph is proposed; based on the probability distribution of the precursor events, pruning is performed on the event logic graph; considering the large event set that occurs along with the events, event prediction is performed using the event logic graph; effectively improving the accuracy of border event prediction.

[0138] Each embodiment in this specification is described in a progressive manner. The key point of each embodiment is to illustrate the differences from other embodiments. The same and similar parts among the embodiments can be referred to each other.

[0139] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A border event prediction method based on event graph, characterized in that: The steps include: S1. Establish a graph of events based on historical event clusters, and calculate the prior probability of the occurrence of subsequent event types corresponding to each event type in the graph of events; S2. Based on the prior probability of the occurrence of the subsequent event type corresponding to each event type, prune the relationship in the event graph, and obtain the new prior probability of the occurrence of the subsequent event type corresponding to each event type in the event graph after pruning; S3. Optimize the event graph based on the accompanying major events corresponding to each event; S4. Based on the current event, the new prior probability of the subsequent event type corresponding to each event type in the optimized event graph is used to predict the subsequent event type that is about to occur.

2. According to claim 1, a border event prediction method based on event graph is characterized in that: The step S1 specifically includes: S11. Let there be a total of c event sets in the historical event set I, expressed as ; Select the i-th event set from the historical event family I , i=1; create a blank event graph; S12. From the event set Select the jth event , and for events Make a judgment: If there is no event in the event graph The corresponding type, then the event After adding the corresponding type to the event graph, check the event again Subsequent events ; If there is an event in the event graph The corresponding type, then view the event Subsequent events ; S13. Events Make a judgment: If there is no event in the event graph The corresponding type, then the event The corresponding types are added to the event graph, and the events in the event graph are Types and events The type of relationship is established, and the corresponding contact value ; If there is an event in the event graph The corresponding type is directly related to the event in the event graph. Types and events The type of relationship is established, and the corresponding contact value ; S14, loop steps S12 to S13 until the current event set is traversed ; S15, let i=i+1, and repeat steps S12-S14 until all historical event sets I are traversed; S16. Take each event type in the event graph as a node, and the subsequent event type corresponding to each event type as a successor node; traverse each node a in the event graph, and check all successor nodes of node a; record all successor nodes of node a as successor node set b, and let the successor node set b have a total of n successor nodes, expressed as ; Get the total value of the connection between node a and the set of subsequent nodes b, expressed as: ； Among them, L a Represents the total value of the connection between node a and the set of subsequent nodes b; b f represents the fth successor node in the successor node set b; Represents node a and subsequent node b f The connection value between them; Get the prior probability of node a and each successor node in the successor node set b, expressed as: ； in, For node a and the subsequent node b f The prior probability is used to indicate that after node a occurs, the subsequent node b occurs f The prior probability of .

3. The border event prediction method based on event graph according to claim 1 is characterized in that: The step S2 specifically includes: S21. Let the event graph contain m nodes, represented as ; If there is a node in the event graph N And its follower node The prior probability between is less than the first threshold, then the node is obtained and rear drive nodes The posterior probability is expressed as: ； in, For Node and rear drive nodes The posterior probability is used to indicate that after the subsequent node After that, the node The posterior probability of Representation Node and rear drive nodes The contact value of Represents the zth node in the event graph N; Representation Node and nodes The contact value of S22, if the node and rear drive nodes If the posterior probability of is less than the second threshold, then the node and rear drive nodes Prune the relationship between S23, repeating steps S21-S22 until all the relationship values ​​corresponding to the prior probabilities less than the first threshold in the event graph N are pruned; S24. Recalculate the prior probability of each node and its successor node, record it as the new prior probability, expressed as: ； in, For Node and rear drive nodes The new prior probability is used to indicate that when a node After that, the rear drive node occurs The prior probability of Representation Node and driver node The contact value of .

4. The border event prediction method based on the event graph according to claim 1 is characterized in that: The step S3 specifically includes: S31. An event composed of multiple events is called a major event. The set of major events is denoted as S. Let there be g major events in total, expressed as ; The major event that occurs after each event is called the accompanying major event, and the accompanying major event cluster corresponding to each node in the event graph N is denoted as NF, and ,in, Represents the set of accompanying large events corresponding to the mth node in the event graph N; The set containing all events is called the complete set E, and there are k events in it, which can be expressed as ; Among them, the currently selected event is , h=1; S32. Count the selected events in the current round All the accompanying major events of ; Screen the selected events from the event map N The corresponding node , and the selection event All accompanying major events are added to the node A collection of accompanying major events middle; S33. Statistics accompanying major events collection The number of occurrences of each major event type in the event tree. If the number of occurrences of a major event type is greater than the preset value, the major event type is used as a node. Frequent itemsets of ; S34. Let h=h+1, repeat the above steps S32~S33, obtain the frequent item set corresponding to each node in the event graph N; and add the frequent item set information to the event graph N.

5. The border event prediction method based on event graph according to claim 4 is characterized in that: In step S32, the statistics of selected events in the current round are The accompanying major events include: Traverse the set of major events S and select the major events that meet the preset conditions as selected events The pre-conditions include: (1) Regions where major events occur and selected events The occurrence area is consistent; (2) Select an event Occurs within the time interval of a major event; or, the time interval between the end of a major event and the selected event The duration is less than the preset duration; (3) Major events do not include selected events .

6. The border event prediction method based on event graph according to claim 1 is characterized in that: The step S4 specifically includes: S41, record the current event as event e, traverse the event graph, and obtain: (1) Node corresponding to event e ; (2) Node The set of rear-drive nodes , let the set of rear-drive nodes There are s rear-drive nodes in ; According to the new prior probability, the set of follower nodes Sort each subsequent node in; (3) Set S of major event types that occurred within a preset time period e ; S42, select the rear drive node set A rear drive node , l=2; S43. Select the rear drive node set The qth follower node in , q=1; S44, the rear drive node With the rear drive node Compare and if l >s, execute step S45; like , then make the following judgment: (1) If the rear drive node There are frequent itemsets belonging to the large event type set S e , then make the following judgment: If the rear drive node There is no frequent item set belonging to the large event type set S e , then the rear drive node Insert to the qth position; let l=l+1, and loop through steps S43-S44; If the successor node has frequent itemsets belonging to the large event type set S e , then judge the new prior probability of event e and the successor node ; if the new prior probability of event e and the successor node is greater than the new prior probability of event e and the successor node , then insert the successor node into the q-th position; let l = l + 1, and loop to execute steps S43 - S44; if the new prior probability of event e and the successor node is less than or equal to the new prior probability of event e and the successor node , then let q = q + 1, if q = l, then let l = l + 1, and loop to execute steps S43 - S44; if q < l, loop to execute step S44; (2) If the rear drive node There is no frequent item set belonging to the large event type set S e , then make the following judgment: If the rear drive node There is a frequent item set belonging to the large event type set S e , then let q = q + 1. If q = l, then let l = l + 1, and loop to execute steps S43 - S44; if q < l, loop to execute step S44; If the successor node does not have a frequent item set belonging to the large event type set S e , then judge the new prior probability of event e and the successor node ; if the new prior probability of event e and the successor node is greater than the new prior probability of event e and the successor node , then insert the successor node into the q-th position; let l = l + 1, and loop to execute steps S43~S44; if the new prior probability of event e and the successor node is less than or equal to the new prior probability of event e and the successor node , then let q = q + 1, if q = l, then let l = l + 1, and loop to execute steps S43~S44; if q < l, loop to execute step S44; S45, end the loop, at this time the rear drive node set All the subsequent nodes in the are sorted according to the predicted probability of occurrence.

Images