Granted Patent B2

US 7,035,849 · App. 10/349,304 · Granted Apr 25, 2006

Abstract

Claims (44)

1. A method of uniquely identifying an object record in a computer database of object records according to a plurality of ranked exact and probabilistic search rules, comprising the steps of:

obtaining application identification information and attributes of a target object;

executing one or more exact-match search rules to search the database of object records for the target object;

executing one or more user defined probabilistic search rules to search the database of object records for the target object if the exact-match search rules retrieve no object record identical to the obtained application identification information and attributes, wherein a list of probable matches to the target object are retrieved and ranked by degree of match probability;

receiving user input of selection of one retrieved object record determined to be the target object record;

updating the database of object records in real time for the selected target object with new attributes and information associated with the target object;

determining an efficiency measure for each of the exact-match and probabilistic search rules according to a plurality of collected statistics for each search rule; where the efficiency measure measures how efficient a corresponding search rule is in finding a match with the target record; and

adjusting a sequence of execution of the exact-match and probabilistic search rules in descending order by efficiency measure,

wherein one of the collected statistcs corresponds to a number of instances that one of the probabilistic search rules retrieves one or more possible matches to the target object record, or to a number of instances that one of the probabilistic search rules retrieves a record previously retrieved by a previously executed search rule, or to a number of instances that one of the probabilistic search rules retieves a record that was not retrieved by a previously executed search rule, or to a number of instances that one of the probabilistic search rules retrieves a plurality of records subsequently determined to correspond to the target object record, or to a number of retrieved records that are determined not to be the target object record.

2. The method of claim 1 , further comprising the step of:

creating a new object record if the exact-match or probabilistic search rules fail to return a record determined by the user to be the target object.

3. The method of claim 1 , wherein the exact-match search rules are executed according to a pre-configured rank order.

4. The method of claim 1 , wherein the probabilistic search rules are executed according to a pre-configured rank order.

5. A computer system tangibly embodied in a computer readable medium for uniquely identifying an object record in a computer database of object records according to a plurality of exact and probabilistic search rules, comprising:

instructions to obtain application identification information and attributes corresponding to a target object contained in the database of object records;

instructions to execute one or more exact-match search rules to search the database of object records for the target object;

instructions to execute one or more user defined probabilistic search rules to search the database of object records for the target object if the exact-match search rules retrieve no object record identical to the obtained application identification information and attributes, wherein a list of probable matches to the target object are retrieved and ranked by degree of match probability;

instructions to receive user input of selection of one retrieved object record determined to be the target object record;

instructions to update the database of object records in real time for the selected target object with new attributes and information associated with the target object;

instructions to determine the efficiency of the exact-match and probabilistic search rules according to a plurality of collected statistics for each search rule; and

instructions to adjust a sequence of execution of the exact-match and probabilistic search rules in descending order by the efficiency,

wherein one of the collected statistics corresponds to a number of instances that one of the probabilistic search rules retrieves one or more records as possible matches to the tares object record, or to a number of instances that one of the probabilistic search rules retrieves a record previously retrieved by a previously executed search rule, or to a number of instances that one of the probabilistic search rules retrieves a record that was not retrieved by a previously executed search rule, or to a number of instances that one of the probabilistic search rules retrieves a plurality of records wherein the plurality of records are subsequently determined to correspond to the target object record, or to a number of records that are determined not to be the target object record.

6. The system of claim 5 , further comprising the step of:

instructions to create a new object record if the exact-match or probabilistic search rules fail to return a record determined by the user to be the target object.

7. The system of claim 5 , wherein the exact-match search rules are executed according to a pre-configured rank order.

8. The system of claim 5 , wherein the probabilistic search rules are executed according to a pre-configured rank order.

9. A rules analyzer method to evaluate and rank search rules for searching a computer database of records, comprising the computer-implemented steps of:

collecting a first plurality of statistical performance values regarding execution of at least one exact-match search rule, wherein said execution searches for a target object record;

collecting a second plurality of statistical performance values regarding execution of each at least one probabilistic search rule;

assigning a priority value for each of the exact-match and probabilistic search rules according to the collected statistical performance values; and

ranking the exact-match and probabilistic search rules in descending order according to the assigned priority values,

wherein one of the second plurality of statistical performance values corresponds to a number of instances where execution of one of the probabilistic search rules retrieves one or more possible matches to the target object record, or to a number of instances where execution of one of the probabilistic search rules retrieves a record previously retrieved by a previously executed search rule, or to a number of instances where execution of one of the probabilistic search rules retrieves a record not retrieved by a previously executed search rule, or to a number of instances where execution of one of the probabilistic search rules retrieves a plurality of records subsequently determined to correspond to the target object record, or to a number of retrieved records that are determined not to be the target object record.

10. The method of claim 9 , further comprising the computer-implemented step of:

determining an efficiency of each exact match search rule according to the collected statistical performance values, wherein the efficiency is a percentage of instances that an exact match search rule returns a possible match upon execution.

11. The method of claim 9 , further comprising the computer-implemented step of: determining a precision of each probabilistic match search rule according to the collected statistical performance values, wherein an efficiency is a percentage of possible matches retrieved by the probabilistic match search rule that are resolved as real matches.

12. A rules analyzer system tangibly embodied in a computer readable medium to evaluate and rank search rules for searching a computer database of records, comprising:

means for collecting a first plurality of statistical performance values regarding execution of at least one exact-match search rule, wherein said execution searches for a target object record;

means for collecting a second plurality of statistical performance values regarding execution of each at least one probabilistic search rule;

means for assigning a priority value for each of the exact-match and probabilistic search rules according to the collected statistical performance values; and

means for ranking the exact-match and probabilistic search rules in descending order according to the assigned priority value,

wherein one of the second plurality of statistical performance values corresponds to a number of instances where execution of one of the probabilistic search rules retrieves one or more possible matches to the target object record, or to a number of instances where execution of one of the probabilistic search rules retrieves a record previously retrieved by a previously executed search rule, or to a number of instances where execution of one of the probabilistic search rules retrieves a record not retrieved by a previously executed search rule, or to a number of instances where execution of one of the probabilistic search rules retrieves a plurality of records subsequently determined to correspond to the target object record, or to a number of records that are determined not to be the target object record.

13. The system of claim 12 , further comprising:

means for determining an efficiency measure of the exact-match search rule according to the collected statistical performance values, wherein the efficiency measure is a percentage of instances that an exact match search rule returns a possible match upon execution; and

means for determining a precision of the probabilistic match search rule according to the collected statistical performance values, wherein an efficiency is a percentage of possible matches retrieved by the probabilistic match search rule that are resolved as real matches.

Assignments (10)