Method of document searching

• US 20050102251A1
• Filed: 12/14/2001
• Published: 05/12/2005
• Est. Priority Date: 12/15/2000
• Status: Abandoned Application

• Current Assignee: David Gillespie
• Original Assignee: David Gillespie
• Inventors: Gillespie, David
• Application Number:
  US10/451,188
• Publication Number:
  US 20050102251A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/3338 Query expansion
• CPC Class Codes: G06F 16/3346 using probabilistic model

Abstract

A method for searching documents, which uses a concept based retrieval methodology, uses for any query an adaptive self-generating neural network for analyzing concepts contained in the documents being searched as such concepts occur. The method automatically creates, abstracts and populates categories of concepts for each query, and is not restricted to language, but can also be applied to non-text data, such as voice, music, image and film. The method is able to deliver search results that are relevant to the query and to the context of the query and, therefore, arranges results by concept, rather than keyword occurrence.

Claims

1-12. -12. (canceled)

13. A method for searching documents by identifying concepts in both a natural language query and in an unstructured data collection, said method comprising the steps of:

matching concepts in a query to concepts in data for locating relevant items in a collection being searched; and

,

clustering the relevant items together into groups wherein members of each group of said groups are conceptually related.

14. The method for searching documents according to claim 13, wherein information retrieval for unstructured data is carried out by an indexing step and a searching step.

15. The method for searching documents according to claim 14, wherein said information retrieval for unstructured data by elements comprising natural language processing, feature extraction, self-generating neural networks and data-clustering.

16. The method for searching documents according to claim 14, further comprising feature extracting wherein said indexing step identifies concepts and constructs an abstract for each item belonging to a collection of said unstructured data.

17. The method for searching documents according to claim 16, wherein said indexing step comprises the step of:

organizing each said item in a manner conductive for concept matching using a self-generating neural network.

18. The method for searching documents according to claim 13, wherein said matching step and said clustering step utilize natural language process, a self-generating neural network and data clustering.

19. The method for searching documents according to claim 13, wherein said matching step includes the steps of:

parsing said query using a natural language processing element;

submitting said query to a self-generating neural network for matching the concepts in said query to said relevant items in said collection; and

,

passing a ranked set of items from said self-generating neural network to a data cluster.

20. The method for searching documents according to claim 19, wherein said data cluster identifies common items from properties of said relevant items, so that said relevant items having similar properties are grouped together for forming a cluster of items.

21. The method for searching documents according to claim 20, further comprising the step of:

generating a label for each said cluster of items representing a property common to all said relevant items in said cluster of items, so that said cluster of items and said relevant items belonging to each said cluster of items are presentable to a user when said method for switching documents is concluded.

22. The method for searching documents according to claim 13, wherein said matching step is performed for non-text material.