System for improving the performance of information retrieval-type tasks by identifying the relations of constituents

• US 7,206,787 B2
• Filed: 10/04/2004
• Issued: 04/17/2007
• Est. Priority Date: 06/18/1999
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Suzuki, Hisami, Corston-Oliver, Simon H., Dolan, William B.
• Primary Examiner(s): Pardo; Thuy N.
• Application Number:
  US10/957,974
• Publication Number:
  US 20050060139A1
• Time in Patent Office: 925 Days
• Field of Search: 707/2, 707/3, 707/10, 707/6, 707/101, 707/102, 707/103.R, 707/104.1, 704/9
• US Class Current: 1/1
• CPC Class Codes: B42D 15/00 Printed matter of special format or style not otherwise provided for
• CPC Class Codes: B42D 15/0053 Forms specially designed for commercial use, e.g. bills, receipts, offer or order sheets, coupons
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: Y10S 707/917 Text
• CPC Class Codes: Y10S 707/954 Relational
• CPC Class Codes: Y10S 707/956 Hierarchical
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation
• 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 707/99942 Manipulating data structure, e.g. compression, compaction, compilation
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface
• CPC Class Codes: Y10S 707/99944 Object-oriented database structure
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A system is utilized for determining a relationship between first and second textual inputs. The system identifies constituents in the first textual input, having predetermined characteristics indicative of usefulness in determining the relationship. The relationship is then determined based on the constituents identified. The constituents can be eliminated from the first textual input, weighted in the first textual input, or simply annotated in one of a variety of ways.

Claims

1. A computer-implemented method of identifying aspects of constituents in a textual input that are useful in determining a meaning of a textual input, the aspects comprising relation or case information, the method comprising:

(a) selecting a first aspect;

(b) identifying constituents in a textual corpus corresponding to the selected aspect; and

(c) obtaining an indication of the usefulness of the identified constituents in determining the meaning of the textual input, wherein selecting a first aspect (c) comprises selecting a first relation;

(d) repeating steps (a)–

(c) for each desired relation; and

(e) ranking the desired relations in a relation hierarchy based on the indication of usefulness for each desired relation.

2. The method of claim 1 wherein the obtaining step (c) comprises:

(c)(1) indexing constituents in a textual corpus having the first relation;

(c)(2) performing information retrieval operations on the index; and

(c)(3) determining performance of the information retrieval operation.

3. The method of claim 2 wherein the ranking step (e) comprises:

(e)(1) ranking desired relations having higher performance higher in the relation hierarchy.

4. A computer-implemented method of identifying aspects of constituents in a textual input that are useful in determining a meaning of a textual input, the aspects comprising relation or case information, the method comprising:

(a) selecting a first aspect;

(b) identifying constituents in a textual corpus corresponding to the selected aspect; and

(c) obtaining an indication of the usefulness of the identified constituents in determining the meaning of the textual input, wherein the selected aspect comprises case information corresponding to constituents in the textual input and wherein identifying constituents comprises identifying constituents in the textual input corresponding to the first case information; and

(d) determining whether to identify the first case information as being useful in determining the meaning of the first textual input based on the indication of usefulness.

5. The method of claim 4 wherein the obtaining step (c) comprises:

(c)(1) removing from the textual input all constituents, having a case indicated by the first case information;

(c)(2) providing an index corresponding to the textual input with the constituents removed;

(c)(3) performing information retrieval operations on the index;

(c)(4) determining a reduction in a size of the index achieved by removing the constituents; and

(c)(5) determining performance of the information retrieval operations.

6. The method of claim 5 wherein the determining step (d) comprises:

(d)(1) if performance is adequate, given the reduction in the size of the index, identifying the selected case information.

7. The method of claim 4 and further comprising:

if the selected case information is not useful, conducting failure analysis to determine whether the selected case information should be conditionally identified as useful.

8. The method of claim 4 and further comprising:

performing steps (a)–

(d) a plurality of times; and

placing the selected case information in a ranked list based on the corresponding indication of usefulness.

9. A computer-implemented method of identifying aspects of constituents in a textual input that are useful in determining a meaning of a textual input, the aspects comprising relation or case information, the method comprising:

(a) selecting a first aspect;

(b) identifying constituents in a textual corpus corresponding to the selected aspect; and

(c) obtaining an indication of the usefulness of the identified constituents in determining the meaning of the textual input;

(d) repeating steps (a)–

(c) for each desired aspect; and

(e) ranking the desired aspect in an aspect hierarchy based on the indication of usefulness of each desired aspect.

10. The method of claim 9 wherein the aspect hierarchy comprises a relation hierarchy and further comprising:

(e) identifying a hierarchy threshold in the relation hierarchy corresponding to the textual input.

11. The method of claim 10 wherein identifying a hierarchy threshold (f) comprises:

f(1) selecting a hierarchy threshold;

f(2) identifying constituents in the textual input having corresponding relations below the hierarchy threshold;

f(3) obtaining an indication of the usefulness of the constituents, other than the identified constituents, in determining the meaning of the textual input; and

f(4) determining whether the selected hierarchy threshold is a suitable threshold based on the indication of usefulness.

12. The method of claim 11 wherein the obtaining step f(3) comprises:

f(3)(i) removing from the textual input all constituents, having a corresponding relation located on the relation hierarchy below the selected hierarchy threshold;

f(3)(ii) providing an index corresponding to the textual input with the constituents removed;

f(3)(iii) performing information retrieval operations on the index;

f(3)(iv)determining a reduction in a size of the index achieved by removing the constituents; and

f(3)(v) determining performance of the information retrieval operations.

13. The method of claim 12 wherein the determining step f(4) comprises:

f(4)(i) if performance is adequate, given the reduction in the size of the index, identifying the selected hierarchy threshold as a suitable threshold.

14. The method of claim 12 wherein the determining performance step f(3)(v) comprises:

determining at least one of speed and accuracy of the information retrieval operations.

15. The method of claim 11 and further comprising:

repeating steps f(1)–

f(4) until a highest suitable threshold has been selected.

16. The method of claim 11 and further comprising:

if the hierarchy threshold is unsuitable, conducting failure analysis to determine whether the selected hierarchy threshold should be conditionally identified as a suitable threshold.

17. The method of claim 11 and further comprising performing steps (a) to (d) in the order listed.