Process and system for recursive document retrieval

• US 5,787,424 A
• Filed: 11/30/1995
• Issued: 07/28/1998
• Est. Priority Date: 11/30/1995
• Status: Expired due to Term

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Electronic Data Systems Corporation (Perspecta, Inc.)
• Inventors: Thompson, Gregory J., Hill, Joe R.
• Primary Examiner(s): Black, Thomas G.
• Assistant Examiner(s): Jung, David Yiuk
• Application Number:
  US08/565,535
• Time in Patent Office: 971 Days
• Field of Search: 395/600, 395/601-623, 707/1-7, 707/100-101, 707/530, 707/531, 707/536, 704/1, 704/2, 704/231, 364/578, 364/724.17, 364/724.2, 364/725.01
• US Class Current: 1/1
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access
• CPC Class Codes: Y10S 707/99936 Pattern matching access
• CPC Class Codes: Y10S 707/99937 Sorting

Abstract

A system (10) for recursive document retrieval is provided. An original repository (12) comprises a plurality of documents (14), and a query (16) comprises information by which the original repository (12) can be analyzed to identify relevant documents (14). A retrieval engine (18) is coupled to the original repository (12) and receives the query (16). The retrieval engine (18) is operable to execute an automatic recursive retrieval process with respect to the original repository (12) using the query (16) where the automatic recursive retrieval process comprises a plurality of iterations. The retrieval engine (18) is also operable to provide as system output a retrieved repository (28) comprising a plurality of documents (30) from the original repository (12) identified as sufficiently relevant to the query (16) according to the automatic recursive retrieval process for detection as relevant documents.

Claims

1. A system for recursive document retrieval, comprising:

an original repository comprising a plurality of documents;

a retrieval engine coupled to the original repository and receiving a query, the retrieval engine operable;

to execute an automatic recursive retrieval process with respect to the original repository using the query, the automatic recursive retrieval process comprising a plurality of iterations; and

to provide as system output a retrieved repository comprising a plurality of documents from the original repository identified as sufficiently relevant to the query according to the automatic recursive retrieval process for detection as relevant documents;

wherein the plurality of iterations comprises a first iteration performing a retrieval process on the original repository to generate an iterative repository and subsequent iterations performing retrieval processes on iterative repositories generated by previous iterations to generate additional iterative repositories; and

wherein the retrieved repository is one of the additional iterative repositories.

2. The system of claim 1, wherein the automatic retrieval process comprises an iteration that uses a Bayesian retrieval process.

3. The system of claim 2, wherein the Bayesian retrieval process uses a relevance measure between documents based upon a log-likelihood ratio generated by assuming that if the document is not relevant, the parameters of a sampling distribution can be drawn from an underlying distribution and if the document is relevant, the parameters of the sampling distribution can be drawn from an updated distribution.

4. The system of claim 3, wherein the original repository comprises documents that each comprise text information and the query comprises text information and wherein the parameters of the sampling distribution comprise word rates.

5. The system of claim 1, wherein the automatic recursive retrieval process comprises iterations that use different types of retrieval processes.

6. The system of claim 1, wherein the system is implemented using a computer system having a storage device, memory and a processor.

7. A process for recursive document retrieval, comprising:

providing an original repository comprising a plurality of documents;

executing an automatic recursive retrieval process with respect to the original repository using a query, the automatic recursive retrieval process comprising a plurality of iterations including;

performing, in a first iteration, a retrieval process on the original repository to generate an iterative repository; and

performing, in subsequent iterations, retrieval processes on iterative repositories generated by previous iterations to generate additional iterative repositories; and

providing as system output a retrieved repository comprising a plurality of documents from the original repository identified as sufficiently relevant to the query for detection as relevant documents, wherein the retrieved repository is one of the additional iterative repositories.

8. The process of claim 7, wherein executing an automatic recursive retrieval process comprises executing an iteration that uses a Bayesian retrieval process.

9. The process of claim 8, wherein executing an iteration that uses a Bayesian retrieval process comprises performing a Bayesian retrieval process that uses a relevance measure between documents based upon a log-likelihood ratio generated by assuming that if the document is not relevant, the parameters of a sampling distribution can be drawn from an underlying distribution and if the document is relevant, the parameters of the sampling distribution can be drawn from an updated distribution.

10. The process of claim 9, wherein providing an original repository comprises providing an original repository comprising documents that each comprise text information and wherein providing a query comprises providing a query comprising text information and wherein the parameters of the sampling distribution comprise word rates.

11. The process of claim 7, wherein executing an automatic recursive retrieval process comprises performing iterations that use different types of retrieval processes.

12. The process of claim 7, wherein the process is accomplished using a computer system having a storage device, memory and a processor.

13. A process for constructing an automatic recursive document retrieval system, comprising:

providing a representation of an original document repository;

providing a representation of a plurality of retrieval processes;

linking a first selected retrieval process from the plurality of retrieval processes to the original repository;

getting parameters for the first retrieval process;

creating a representation of a first iterative repository produced by the first retrieval process; and

repeating linking, getting and creating for additional selected retrieval processes and iterative repositories in order to construct a model of the automatic recursive retrieval system.

14. The process of claim 13, further comprising implementing the automatic recursive retrieval system based upon the model.

15. The process of claim 14, further comprising receiving a query and executing the automatic recursive retrieval system to identify and provide retrieved documents as process output.

16. The process of claim 13, wherein providing a representation of a plurality of retrieval processes comprises providing a representation of a Bayesian retrieval process.

17. The process of claim 16, wherein providing a representation of a Bayesian retrieval process comprises providing a representation of a Bayesian retrieval process that uses a relevance measure between documents based upon a log-likelihood ratio generated by assuming that if the document is not relevant, the parameters of a sampling distribution can be drawn from an underlying distribution and if the document is relevant, the parameters of the sampling distribution can be drawn from an updated distribution.

18. The process of claim 17, wherein the parameters of the sampling distribution comprise word rates.

19. The process of claim 13, wherein the process is accomplished using a computer system having a storage device, memory and a processor.