MODEL-DRIVEN EVENT DETECTION, IMPLICATION, AND REPORTING SYSTEM

• US 20090192965A1
• Filed: 03/16/2009
• Published: 07/30/2009
• Est. Priority Date: 04/08/2005
• Status: Active Grant

• Current Assignee: Accenture Global Services GmbH (Accenture PLC)
• Original Assignee: Accenture Global Services GmbH (Accenture PLC)
• Inventors: KASS, Alex, Hughes, Lucian P., Sheth, Neil, Cowell-Shah, Christopher
• Granted Patent: US 8,538,911 B2
• Time in Patent Office: Days
• US Class Current: 706/46
• CPC Class Codes: G06F 16/2465 Query processing support for facilitating data mining operations in structured databases
• CPC Class Codes: G06Q 10/00 Administration; Management
• CPC Class Codes: Y10S 707/99942 Manipulating data structure, e.g. compression, compaction, compilation
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing
• CPC Class Codes: Y10S 707/99948 Application of database or data structure, e.g. distributed, multimedia, or image

Abstract

An event analysis system monitors information available from both publicly and privately distributed networks of information for events that are relevant to the user'"'"'s particular business concerns. Those concerns are defined in a customized model of the user'"'"'s organization and external business environment. The system receives the information, detects events in the information, interprets the events, and determines implications of these events. The detection and implication proceeds with regard to specific entities, relationships between entities, and definitions of the types of events which may occur in the environment in which the entities exist. Accordingly, the analysis system intelligently adapts its processing to recognize and report events which may be of interest for any particular entity.

Claims

1. An event analysis system comprising:

an information database;

a memory comprising;

an event model comprising an event type and event attributes of the event type;

an event detection engine;

an event implication engine;

an environment model comprising a first model entity and a focus entity, and a focus relationship from the focus entity to the first model entity; and

a processor coupled to the memory which is operable to;

initiate execution of the event detection engine on articles to detect an event involving the first model entity represented in the articles and apply the event model to generate an event object according to a common event structure;

initiate execution of the event implication engine on the event object to;

recognize that the focus relationship exists between the focus entity and the first model entity;

determine an inferred event on behalf of the focus entity in view of the focus relationship responsive to the detection of the event involving the first model entity; and

create a new event object from the inferred event.

2. The event analysis system of claim 1,

the environment model further comprising fill entities and a fill relationship from the fill entities to the first model entity, the fill entities specifying alternative focus entities;

and

the processor further operable to initiate execution of the event implication engine on the event object to determine multiple inferred events on behalf of the fill entities in view of the fill relationship, where the event implication engine recognizes that the fill relationship exists between the fill entities and the first model entity and responsively generates the multiple inferred events due to detection of the event involving the first model entity, and creates multiple new event objects from the multiple inferred events.

3. The event analysis system of claim 1,

the event object comprising:

an event type field;

an event type probability field;

an importance field; and

a public interest field;

an network interface operable to receive a public interest level value and importance level value of the event; and

where the event detection engine is configured to;

store an event type value in the event type field that indicates an event type;

store an event type probability field value in the probability field that indicates the probably that the event type has occurred;

store the importance level value in the importance field that indicates the importance level of the event; and

store the public interest level value in the public interest field that indicates the level of interest of the public in the event.

4. The event analysis system of claim 1, where the event detection engine is further configured to:

determine whether any entities in the environment model are associated with the event; and

discard the event when no entity in the environment model is associated with the event.

5. The event analysis system of claim 1, where the memory further comprises:

an event implication model comprising a trigger constraint and a resulting implication, and where the event implication engine applies the event implication model to determine the resulting implication.

6. The event analysis system of claim 1, where the processor is further operable to:

apply the event implication engine to the new event object.

7. The event analysis system of claim 3, where the information database further comprises:

an information source model that comprises reputation, reliability and quality characteristics of a first information source;

additional articles from a second information source; and

where the information source model further specifies the second information source, including reputation, reliability and quality characteristics of the second information source; and

where the event detection engine is further operable to;

resolve conflicts between articles obtained from the first information source and the second information source by analyzing the reputation, reliability and quality characteristics in the information source model.

8. The event analysis system of claim 1, further comprising:

an event database which stores the event object.

9. The event analysis system of claim 4, where the event detection engine is operable to:

apply a filter to determine whether the event object specifies an entity in the environment model.

10. The event analysis system of claim 9, where the event detection engine is further operable to:

apply a classification algorithm to the event object to determine the event type represented in the event object.

11. The event analysis system of claim 10, where the event detection engine is further operable to:

extract attributes specific to the event type from article text in the event object; and

store the attributes into attribute fields of the event object;

12. The event analysis system of claim 11, where the event object includes an event description, and where the event detection engine is operable to:

tokenize the event description to provide a tokenized description; and

tag tokens in the tokenized description with parts-of-speech to provide a tagged description.

13. The event analysis system of claim 12, where the event detection engine is further operable to:

perform named entity recognition on the tagged description;

parse the tagged description to identify grammatical structure; and

perform template matching to determine event attribute values for the event type.

14. The event analysis system of claim 3, where the memory further comprises an event modeling tool operable to display a user interface for modifying the event model.

15. The event analysis system of claim 4, where the memory further comprises an environment modeling tool operable to display a user interface for modifying the environment model.

16. The event analysis system of claim 5, where the memory further comprises an implication modeling tool operable to display a user interface for modifying the event implication model.

17. A computer implemented method for event analysis comprising:

reading from a computer readable memory an information source model to determine an information source;

retrieving an article from the information source;

reading from a computer readable memory an environment model comprising a first model entity and a focus entity, and a focus relationship from the focus entity to the first model entity;

initiating, with a processor coupled to the computer readable memory, execution of an event detection engine on the article to detect an event involving the first model entity represented in the article which is relevant to the focus entity based on the focus relationship and generate an event object;

reading from a computer readable memory an event implication model;

initiating execution, with a processor, of an event implication engine on the event object to;

recognize that the focus relationship exists between the focus entity and the first model entity;

determine an inferred event on behalf of the focus entity in view of the focus relationship responsive to the detection of the event involving the first model entity; and

creating a new event object from the inferred event.

18. The computer implemented method for event analysis of claim 17, further comprising:

reading, from the environment model, fill entities and a fill relationship from the fill entities to the first model entity, the fill entities specifying alternative focus entities; and

initiating execution of the event implication engine on the event object to determine multiple inferred events on behalf of the fill entities in view of the fill relationship; and

recognizing that the fill relationship exists between the fill entities and the first model entity and responsively generates the multiple inferred events due to detection of the event involving the first model entity, and creates multiple new event objects from the multiple inferred events.

19. The method of claim 17, where reading an event implication model comprises:

reading a trigger constraint and a resulting implication.

20. The method of claim 19, where reading a trigger constraint comprises:

reading an event constraint and an attribute constraint on the event constraint.

21. The method of claim 17, the event object comprising:

an event type field;

an event type probability field;

an importance field; and

a public interest field.

22. The method of claim 17, further comprising:

storing the new event object in an event database; and

initiating execution of the event implication engine on the new event object.

23. The method of claim 17, further comprising:

determining an event type based on the article;

creating the event object of the event type, including an event description from the article; and

storing the event object in an event database.

24. The method of claim 23, where initiating execution of the event detection engine comprises:

initiating execution of a tokenizer on the event description to generate a tokenized description; and

tagging tokens in the tokenized description to provide a tagged description.

25. The method of claim 24, further comprising:

performing named entity recognition on the tagged description; and

performing template matching to determine event attribute values for the event type.

26. The method of claim 23, further comprising:

applying a classification algorithm to the event object to determine the event type represented in the event object.

27. The method of claim 26, further comprising:

reading an event model to determine event attributes for the event type; and

extracting values for the event attributes from the event object.

28. An event analysis product comprising:

a machine-readable medium;

event processing control instructions stored on the machine-readable medium operable to;

read an environment model comprising a first model entity and a focus entity;

read an information source model to determine information sources;

retrieve articles from the information sources and store the articles in an information database;

initiate execution of an event detection engine on articles to detect events represented in the articles, generate event objects according to a common event structure, and store the event objects in an event database;

initiate execution of an event implication engine on the event objects to;

recognize that the focus relationship exists between the focus entity and the first model entity;

determine an inferred event on behalf of the focus entity in view of a focus relationship responsive to the detection of the event involving the first model entity, generate a new event object from the inferred event, and store the new event object in the event database; and

initiate communication of the event objects including the inferred event to an external system subscriber to the event analysis product.

29. The event analysis product of claim 28, the instructions further operable to:

read fill entities and a fill relationship from the fill entities to the first model entity, wherein the environment model comprises the fill entities and the fill relationship, the fill entities specifying alternative focus entities; and

initiate execution of the event implication engine on the event object to determine multiple inferred events on behalf of the fill entities in view of the fill relationship, where the event implication engine recognizes that the fill relationship exists between the fill entities and the first model entity and responsively generates the multiple inferred events due to detection of the event involving the first model entity, and creates multiple new event objects from the multiple inferred events.

30. The event analysis product of claim 28, the instructions further operable to publish the event object to a message subscription/publication service.

31. The event analysis product of claim 28, the instructions further operable to:

re-initiate execution of the event implication engine on the new event object, generate an additional event object, and store the additional event object in the event database, and

receive a public interest level value and importance level value for each event, where the event model comprises an event type and event attributes of the event type, and where the new event object and the additional event object each comprise;

an event type field;

an event type probability field;

an importance field; and

a public interest field.

32. The event analysis product of claim 28, the instructions further operable to read an event implication model comprising a trigger constraint and a resulting implication.

33. The event analysis product of claim 32, where the trigger constraint comprises an event constraint and an attribute constraint on the event constraint.

34. The event analysis product of claim 28, where the environment model comprises the focus relationship specifying a relationship between the focus entity and the first model entity.

35. The event analysis product of claim 34, the instructions further operable to:

determine event entities including the first model entity associated with the events; and

filter the events based on presence of the focus entity in the focus relationship.