Temporal Document Trainer and Method

• US 20090048990A1
• Filed: 08/14/2008
• Published: 02/19/2009
• Est. Priority Date: 08/14/2007
• Status: Active Grant

• Current Assignee: John Nicholas and Kristin Gross Trust U/A/D April 13 2010
• Original Assignee: John Nicholas and Kristin Gross Trust U/A/D April 13 2010
• Inventors: Gross, John Nicholas
• Granted Patent: US 8,442,923 B2
• Time in Patent Office: Days
• US Class Current: 706/12
• CPC Class Codes: G06F 16/24 Querying
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/95 Retrieval from the web
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/9536 Search customisation based on social or collaborative filtering
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06Q 30/00 Commerce
• CPC Class Codes: G06Q 30/0243 Comparative campaigns
• CPC Class Codes: G06Q 30/0251 Targeted advertisements

Abstract

An electronic document sorter is trained to classify documents based on their temporal qualities. The invention can be used in environments such as automated news aggregators, search engines and other electronic systems which compile information having temporal qualities.

Claims

1. A method of automatically training an electronic document sorter with a computing system comprising:

a) identifying a first set of set of electronic documents containing content associated with a set of topics using the computing system;

wherein said content describes one or more events related to one or more topics;

b) analyzing said first set of electronic documents to identify first temporal components for each of said one or more events and one or more topics;

wherein said temporal components include content other than but not excluding date/time data specified for said first document.

c) training an automated document classifier using said temporal components;

wherein the automated document classifier is adapted by such training to classify an unclassified electronic document containing content associated with one or more of said topics based on the temporal characteristics of such electronic document.

2. The method of claim 1 wherein said unclassified electronic document is given at least one of a temporal score or a temporal rank.

3. The method of claim 1 wherein said automated document classifier further is trained to determine an event associated with an unclassified electronic document, and receives temporal component information from a reference document set containing reference temporal data corresponding to predicted content for said event.

4. The method of claim 3 wherein said predicted content includes one or more objects and related object states for said predicted content.

5. The method of claim 1 wherein said temporal components are derived from annotations and/or codings made to such first set of electronic documents provided by a human reviewer.

6. The method of claim 1 wherein said temporal components are derived from annotations and/or codings made automatically to said first set of electronic documents by a text parser.

7. The method of claim 1 wherein said automated document classifier is adapted to use semantic data for at least some of said content associated with said unclassified electronic document.

8. The method of claim 1 further including a step:

tuning said automated document classifier with additional temporal related data for a class of events derived from a subsequent set of electronic documents relating to such class of events.

9. The method of claim 1 wherein said events are sports related, natural disaster related, accident related, political election related, and/or financial market related.

10. The method of claim 1 wherein said automated document classifier is used with an automated news aggregator executing on a web site.

11. The method of claim 1 wherein said automated document classifier is used with a search engine.

12. The method of claim 1 wherein said automated document classifier is used with a recommender system.

13. The method of claim 1 wherein said temporal components are stored as selected content depicting separate sequential temporal states of said event.

14. The method of claim 13, wherein said selected content also includes variable data designed to accommodate data which indirectly indicates a temporal state.

15. The method of claim 13 wherein said variable data relates to one of at least a sporting event score, a number of persons in a disaster or accident, an election result, a movie box office total, a number of shares traded, a medical condition, and time dependent physical variables.

16. The method of claim 1 wherein said electronic documents are based on transcriptions of speech recognized utterances.

17. The method of claim 1, wherein step (b) is performed by analyzing browsing behavior of users reviewing said first set of electronic documents.

18. The method of claim 1 wherein said steps are implemented as one or more computer software routines embodied in a tangible media and adapted to cause one or more computing systems to perform the operations recited therein.

19. The method of claim 1 wherein automated document classifier generates a temporal template for each of said one or more topics, such that a plurality of temporal states are mapped to a plurality of respective content items.

20. The method of claim 1 wherein said electronic documents are based on a transcription of an audio/video file and/or metadata associated therewith.

21. The method of claim 1 wherein an author and/or a locale associated with said documents are also identified and classified with said documents.

22. The method of claim 1 where said first temporal components for each of said one or more events and one or more topics are identified by classifying content differences between documents.

23. A method of automatically training an electronic document sorter with a computing system comprising:

a. identifying a first set of set of electronic documents containing content associated with a set of topics using the computing system;

wherein said content describes one or more events related to one or more topics;

b. sorting said first of electronic documents in temporal order with respect to each of said one or more events and one or more topics to form at least one sorted topic/event document set using the computing system;

c. analyzing said at least one sorted topic/event document set using the computing system to identify temporal content relating to each electronic document which correlates to a temporal rank of such electronic document;

d. training an automated document classifier using the computing system so that it can automatically identify such temporal components in other electronic documents associated with a particular event/topic and which are not part of said sorted topic/event document set.

24. The method of claim 23 wherein said temporal components include semantic equivalents.

25. The method of claim 23 wherein an author and/or a locale associated with said documents are also identified and classified with said documents.

26. The method of claim 23 where said temporal component are identified by classifying content differences between documents.

27. The method of claim 23 wherein said steps are implemented as one or more computer software routines embodied in a tangible media and adapted to cause one or more computing systems to perform the operations recited therein.