Data mining by retrieving causally-related documents not individually satisfying search criteria used

• US 7,386,439 B1
• Filed: 02/04/2003
• Issued: 06/10/2008
• Est. Priority Date: 02/04/2002
• Status: Active Grant

• Current Assignee: Ernst & Young US LLP (Ernst & Young LLP)
• Original Assignee: Cataphora Incorporated
• Inventors: Holsinger, David J., Brouk, Roman V., Roberts, Steven L., Charnock, Elizabeth
• Primary Examiner(s): {hacek over (S)}mits; Talivaldis Ivars
• Application Number:
  US10/358,793
• Time in Patent Office: 1,953 Days
• Field of Search: 704/3, 704/4, 715/530
• US Class Current: 704/9
• CPC Class Codes: G06F 16/335 Filtering based on additional data, e.g. user or group profiles filtering in web context G06F16/9535, G06F16/9536
• CPC Class Codes: Y10S 707/99931 Database or file accessing

Abstract

This patent describes a method and apparatus to automatically and accurately winnow down arbitrarily large amounts of electronic information created by a particular population of actors to only those subsets of particular interest by having a causal relationship, even when retrieved documents containing this information do not individually satisfy the search criteria used. An actor in this context is defined as any entity, single or aggregate, capable of creating, distributing, modifying, or receiving digital information. Once identified, this subset of information may, for example, be processed, analyzed, redacted, or destroyed, depending on the context of the system'"'"'s use.

Claims

1. A computer-aided method of providing analysis in a corpus of documents comprising at least one discussion, the method comprising:

analyzing the discussion to identify a communicative intent of an author of the discussion, including topic, an intended audience, and beginning and ending of a series of communications comprising the discussion even when the retrieved documents do not individually satisfy the search criteria used, the analysis providing a causal relationship between the series of communications.

2. The method of claim 1, wherein the discussion includes a plurality of text blocks, and further comprising:

attributing each of the plurality of text blocks to its respective author.

3. The method of claim 2, further comprising:

identifying potentially damaging content within the discussion, and tagging the author contributing to the potentially damaging content.

4. The method of claim 3, further comprising:

tracking actor communications throughout the corpus of documents in which an actor has contributed to the potentially damaging content.

5. The method of claim 4, wherein the tracking further includes an author who has viewed the potentially damaging content.

6. The method of claim 3, wherein the potentially damaging content is used to identify patterns of behavior.

7. The method of claim 1, further comprising:

evaluating completeness of a corpus of documents, by comparing at least two overlapping datasets; and

identifying systematic deletions.

8. The method of claim 7, further comprising:

identifying an actor'"'"'s lifespan; and

comparing a period of analysis associated with the corpus with the actor'"'"'s lifespan to identify troughs in the documents when according to the lifespan the actor was present, but no corresponding documents are present in the corpus.

9. The method of claim 1, further comprising identifying content as being of special interest, and flagging a document including the content and an actor that created the content.

10. The method of claim 1, further comprising:

detecting a pattern of communication between actors; and

flagging an unexplained trough in the pattern of communications.

11. The method of claim 1, further comprising:

comparing a list of tagged actors to the actor “

circles of trust”

; and

detecting and flagging the appearance of more than one actor from the same circle of trust in the tagged list to detect conspiracies and find anomalies such as the systematic deletion of discussions among certain sets of actors.

12. The method of claim 1, wherein the document is captured based on an imprecise input technique, the imprecise input techniques including optical character recognition (OCR).

13. The method of claim 12, further comprising:

generating an actor graph; and

when the system identifies an anomalous name resulting from character corruption during the imprecise input, utilizing the actor graph to evaluate a probability of a match with an existing name.

14. The method of claim 13, further comprising:

removing a spurious actor from the actor graph, when the analysis determines that the spurious actor is a misidentification of another actor, and ensuring that the documents attributed to the spurious actor are reattributed.

15. A computer-aided method comprising:

analyzing a discussion to identify a communicative intent of an author of the discussion, including topic, an intended audience, a beginning, and ending of a series of communications comprising the discussion even when the documents retrieved therefor do not individually satisfy the search criteria used; and

determining anomalies in a corpus of documents, wherein the corpus of documents includes multiple data sets from multiple sources, and the anomalies represent substantive differences between the multiple data sets.

16. The method of claim 15, wherein determining anomalies comprises:

analyzing of a set of overlapping documents.

17. The method of claim 16, further comprising:

calculating actor graphs including a circle of trust.

18. The method of claim 17, further comprising:

ranking proximity between actors in the multiple data sets.

19. The method of claim 18, wherein proximity is a function of a number of links two actors are apart from each other.

20. The method of claim 15, further comprising:

determining an amount of text associated with an actor based on topic.

21. The method of claim 20, further comprising:

comparing a percentage of material created by the author to a percentage of the material received by the author on a particular topic.

22. The method of claim 15, further comprising:

identifying overlapping portions of a plurality of data sets; and

comparing the overlapping portions of the plurality of data sets to compute a difference between the data sets.

23. The method of claim 22, wherein the difference comprises a degree of absolute and relative difference among the data sets.

24. The method of claim 22, wherein deviation from what is considered customary behavior is examined within a particular given context.

25. The method of claim 22, further comprising:

enabling evaluation of the statistical analysis based on actor and/or topic.

26. A computer-aided method comprising:

analyzing a discussion comprising a plurality of causally related documents to identify a communicative intent of an author of the discussion, including topic, an intended audience, a beginning, and ending of a series of communications comprising the discussion even when the retrieved documents do not individually satisfy the search criteria used; and

enabling redaction of a portion of the document, by removing data from a document and replacing the removed data it with data of no value.

27. The method of claim 26, wherein the redaction includes text blocks immediately following, when the immediately following text blocks are no more than one sentence long, when the immediately following text blocks are a near match with the redacted portion of the document, to remove hints regarding the redacted text.

28. The method of claim 26, further comprising:

when the text block includes text to be redacted and text that must be kept, dividing sentences into a plurality of sub-clauses; and

separately analyzing each sub-clause and sentence for redaction.