Generalized data mining and analytics apparatuses, methods and systems

• US 9,183,203 B1
• Filed: 10/04/2011
• Issued: 11/10/2015
• Est. Priority Date: 07/01/2009
• Status: Active Grant

• Current Assignee: Quantifind Incorporated
• Original Assignee: Quantifind Incorporated
• Inventors: Tuchman, Ari, Galant, Yaron, Nachbar, Erich, Stockton, John, Thiyagarajan, Karthik
• Primary Examiner(s): Fan, Shiow-Jy
• Application Number:
  US13/252,559
• Time in Patent Office: 1,498 Days
• Field of Search: 707/640, 707/609, 707/791, 707/802, 707/821, 707/759, 707/713, 707/769, 707/999.005
• US Class Current: 1/1
• CPC Class Codes: G06F 16/30 of unstructured textual data document management systems G06F16/93
• CPC Class Codes: G06F 16/3322 using system suggestions G06F16/3325 takes precedence
• CPC Class Codes: G06F 16/3325 Reformulation based on results of preceding query
• CPC Class Codes: G06F 16/3328 using graphical result space presentation or visualisation
• CPC Class Codes: G06F 16/3347 using vector based model
• CPC Class Codes: G06F 16/338 Presentation of query results
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 2216/11 Patent retrieval

Abstract

The GENERALIZED DATA MINING AND ANALYTICS APPARATUSES, METHODS AND SYSTEMS (“GDMA”), in various embodiments, may identify statistical relationships among query terms by analyzing a corpus of electronic documents. Inputs may be automatically generated automatically and/or user provided. In one embodiment, a method includes: accessing a term tensor associated with at least one term in a corpus of documents, wherein the term tensor comprises a plurality of data type vectors corresponding respectively to a plurality of term-correlated data types correlated with the at least one term in the corpus and each data type vector comprising a plurality of binned data type values with corresponding weighted occurrence values derived from the corpus; providing at least one of the plurality of term-correlated data types for selectable display; receiving at least one term-correlated data type selection; and providing data type values associated with the at least one term-correlated data type selection for display.

Claims

1. A data extraction processor-implemented method, comprising:

accessing, via a processor, a term tensor associated with at least one term in a corpus of documents,

wherein the term tensor comprises a plurality of data type vectors corresponding respectively to a plurality of term-correlated data types contextually correlated with the at least one term in the corpus based at least in part on distances between the at least one term and each of the term-correlated data types in the documents, and

each data type vector comprising a plurality of binned data type values, the plurality of binned data type values comprising a discrete representation of term-correlated data type data associated with the each data type vector and each binned value of the plurality of binned data type values having a corresponding weighted occurrence value derived from the corpus;

providing, via the processor, at least one of the plurality of term-correlated data types for selectable display;

receiving, via the processor, at least one term-correlated data type selection; and

providing, via the processor, a subset of the plurality of binned data type values associated with the at least one term-correlated data type selection for display.

2. The method of claim 1, further comprising:

receiving the corpus of documents.

3. The method of claim 1, wherein the at least one term is a query term.

4. The method of claim 3, further comprising:

receiving the query term from a user.

5. The method of claim 3, further comprising:

identifying corpus terms having frequency or value different from a historical corpus baseline; and

determining the query term based on identified corpus terms.

6. The method of claim 1, wherein accessing a term tensor further comprises:

querying at least one inverted index based on the at least one term to retrieve term tensor data.

7. The method of claim 6, further comprising:

identifying a query type associated with the at least one query term; and

selecting the at least one inverted index from a plurality of inverted indices based on the query type.

8. The method of claim 1, wherein the at least one query term is a keyword, and the at least one term-correlated data type selection is a unit.

9. The method of claim 8, wherein the plurality of binned data type values comprise discrete numerical values associated with the unit.

10. The method of claim 8, wherein the plurality of binned data type values comprise a set of non-numerical values associated with the unit.

11. The method of claim 1, further comprising:

providing weighted occurrence values corresponding to the data type values for display.

12. The method of claim 1, further comprising:

accessing a second plurality of term-correlated data types correlated with the at least one term-correlated data type selection;

providing the second plurality of term-correlated data types for selectable display;

receiving a second at least one term-correlated data type selection, selected from the second plurality of term-correlated data types; and

providing second data type values associated with the second at least one term-correlated data type selection for display.

13. The method of claim 12, wherein the data type values and second data type values are configured as ordered pairs displayed in an XY data representation.

14. The method of claim 1, further comprising:

receiving at least one alert condition;

updating data type values;

evaluating the at least one alert condition based on updated data type values; and

providing an alert if the at least one alert condition is satisfied.

15. The method of claim 1, wherein the at least one term comprises a first term and a second term, corresponding respectively to a first entity and a second entity, and further comprising:

accessing first data type values associated with the at least one term-correlated data type selection and the first term;

accessing second data type values associated with the at least one term-correlated data type selection and the second term;

determining a similarity between the first entity and the second entity based on the first data type values and the second data type values.

16. The method of claim 15, wherein determining a similarity comprises evaluating a sum of products of first data type values and second data type values.

17. The method of claim 15, wherein determining a similarity further comprises:

performing a normalization operation on at least one of the first data type values and the second data type values.

18. The method of claim 1, wherein the at least one term comprises a first term and a second term, corresponding respectively to a first entity and a second entity, and further comprising:

accessing first data types associated with the first term;

accessing second data types associated with the second term; and

determining a similarity between the first entity and the second entity based on an overlap of the first data types and the second data types.

19. The method of claim 1, wherein the weighted occurrence values are determined as a sum of weights for each occurrence of the data type values in the electronic document corpus.

20. The method of claim 1, wherein the plurality of binned data type values and corresponding weighted occurrences are provided for display in a histogram data representation.

21. The method of claim 1, wherein the plurality of binned data type values include both numerical and non-numerical values.

22. The method of claim 1, wherein the plurality of data type values are time-resolved.

23. The method of claim 1, wherein corresponding weighted occurrences are linked to document metadata.

24. A data extraction apparatus, comprising:

a memory;

a processor disposed in communication with the memory and configured to execute a plurality of processing instructions stored in the memory, wherein the processor executes instructions to;

access a term tensor associated with the at least one term in a corpus of documents,

wherein the term tensor comprises a plurality of data type vectors corresponding respectively to a plurality of term-correlated data types contextually correlated with the at least one term in the corpus based at least in part on distances between the at least one term and each of the term-correlated data types in the documents, and

each data type vector comprising a plurality of binned data type values, the plurality of binned data type values comprising a discrete representation of term-correlated data type data associated with the each data type vector and each binned value of the plurality of binned data type values having corresponding weighted occurrence value derived from the corpus;

provide at least one of the plurality of term-correlated data types for selectable display;

receive at least one term-correlated data type selection; and

provide a subset of the plurality of binned data type values associated with the at least one term-correlated data type selection for display.

25. A non-transitory, machine-readable medium, comprising:

a plurality of processing instructions stored in the medium and issuable by a processor to;

access a term tensor associated with the at least one term in a corpus of documents,

wherein the term tensor comprises a plurality of data type vectors corresponding respectively to a plurality of term-correlated data types contextually correlated with the at least one term in the corpus based at least in part on distances between the at least one term and each of the term-correlated data types in the documents, and

each data type vector comprising a plurality of binned data type values, the plurality of binned data type values comprising a discrete representation of term-correlated data type data associated with each data type vector and each binned value of the plurality of binned data type values having corresponding weighted occurrence value derived from the corpus;

provide at least one of the plurality of term-correlated data types for selectable display;

receive at least one term-correlated data type selection; and

provide a subset of the plurality of binned data type values associated with the at least one term-correlated data type selection for display.