Method and apparatus for updating records in a database system based on an improved model of time-dependent behavior

• US 6,549,919 B2
• Filed: 12/08/2000
• Issued: 04/15/2003
• Est. Priority Date: 04/03/2000
• Status: Expired due to Term

• Current Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Original Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Inventors: Sun, Don X., Lambert, Diane, Pinheiro, Jose C.
• Primary Examiner(s): Metjahic, Safet
• Assistant Examiner(s): AL HASHEMI, SANA A
• Application Number:
  US09/733,656
• Publication Number:
  US 20030028502A1
• Time in Patent Office: 858 Days
• Field of Search: 707/203, 707/102, 707/200, 705/60, 710/317, 725/22
• US Class Current: 707/695
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: Y10S 707/99954 Version management

Abstract

A customer signature or other type of record in a database system is updated using an event-driven estimator based on a model of time-dependent behavior. In a preferred embodiment, a current version of a record affected by a given transaction is retrieved from a memory of the system and an event-driven estimator of at least one of a transaction rate and a period probability for the record is determined based at least in part on a dynamic Poisson timing model having a number of periods and corresponding period-based transaction rates associated therewith. The event-driven estimator may be configured so as to generate an estimated transaction rate {circumflex over (λ)}_j,n for period j and a given transaction n, and then to generate an estimated period probability {circumflex over (π)}_j,n for period j as {circumflex over (λ)}_j,n/Σ_k{circumflex over (λ)}_k,n. An updated version of the record may then be generated based on the event-driven estimator.

Claims

1. A method of updating a record in a database system in conjunction with one or more transactions involving at least a portion of the record, the method comprising the steps of:

determining a record affected by a given one of the transactions;

retrieving a current version of the record from a memory of the system;

estimating at least one of a transaction rate and a period probability for the record using an event-driven estimator based at least in part on application of at least a portion of a stochastic model of time-dependent behavior to the current version of the record, the model having associated therewith a plurality of periods and a number of period-based transaction rates; and

generating an updated version of the record based on the event-driven estimator.

2. The method of claim 1 wherein the record comprises a signature associated with a particular customer for which information is stored in the database system.

3. The method of claim 1 wherein the model of time-dependent behavior comprises a constant rate Poisson timing model.

4. The method of claim 1 wherein the model of time-dependent behavior comprises a dynamic Poisson timing model.

5. The method of claim 1 wherein the model of time-dependent behavior comprises a Poisson timing model having deterministic shifts in one or more transaction rates.

6. The method of claim 1 wherein the event-driven estimator for a given one of the periods comprises a weighted average of a previous reciprocal transaction rate and a time since a previous transaction in the given period, and wherein the event-driven estimator for at least a subset of the other periods increases by an amount of time that has been spent in the given period since the previous transaction.

7. The method of claim 1 wherein the event-driven estimator is updated for every transaction in a specified one of the plurality of periods and at the end of each of at least a subset of the plurality of periods.

8. The method of claim 1 wherein the event-driven estimator is updated for every transaction, regardless of which of the plurality of periods any particular transaction falls in, but not at the end of each of the plurality of periods.

9. The method of claim 1 wherein the event-driven estimator is updated for a given one of the plurality of periods only when there is a transaction occurring within that period.

10. The method of claim 1 wherein the event-driven estimator is updated once for a given one of the plurality of periods after a plurality of transactions occurring within that period.

11. The method of claim 1 wherein the event-driven estimator first generates an estimated transaction rate {circumflex over (λ

)}_j,n for period j and a given transaction n, and then generates an estimated period probability {circumflex over (π

)}_j,n for period j as ${\widehat{\mathrm{\lambda <br><br>}}}_{j,n}/\underset{{}_k\widehat{\mathrm{\lambda <br><br>}}_{k,n}}{\sum <br><br>}.$

12. The method of claim 1 wherein an estimated transaction rate {circumflex over (λ

)}_j,n under the model satisfies ${\widehat{\mathrm{\lambda <br><br>}}}_{j,n}^{-1}=\{\begin{array}{cc}\left(1-w\right)<br><br>{\widehat{\mathrm{\lambda <br><br>}}}_{j,n-1}^{-1}+{\mathrm{wZ}}_{j,n,}& \mathrm{if}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{transaction}<br><br>\text{\hspace{1em}<br><br>}<br><br>n<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{falls}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{in}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{period}<br><br>\text{\hspace{1em}<br><br>}<br><br>j\\ {\widehat{\mathrm{\lambda <br><br>}}}_{j,n-1}^{-1}+\frac{w}{1-w}<br><br>Z_{j,n,}& \mathrm{if}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{transaction}<br><br>\text{\hspace{1em}<br><br>}<br><br>n<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{does}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{not}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{fall}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{in}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{period}<br><br>\text{\hspace{1em}<br><br>}<br><br>j,\end{array}$

wherein w denotes a specified updating weight, Z_j,n is an elapsed time in period j since a previous transaction n−

1, and {circumflex over (λ

)}_j,n<sup>−

1</sup> is a reciprocal rate that estimates the average time between transactions in period j.

13. An apparatus for updating a record in a database system in conjunction with one or more transactions involving at least a portion of the record, the apparatus comprising:

a memory for storing at least a portion of the record; and

a processor coupled to the memory and operative to determine a record affected by a given one of the transactions;

to retrieve a current version of the record from a memory of the system;

to estimate at least one of a transaction rate and a period probability for the record using an event-driven estimator based at least in part on application of at least a portion of a stochastic model of time-dependent behavior to the current version of the record, the model having associated therewith a plurality of periods and a number of period-based transaction rates; and

to generate an updated version of the record based on the event-driven estimator.

14. The apparatus of claim 13 wherein the record comprises a signature associated with a particular customer for which information is stored in the database system.

15. The apparatus of claim 13 wherein the model of time-dependent behavior comprises a constant rate Poisson timing model.

16. The apparatus of claim 13 wherein the model of time-dependent behavior comprises a dynamic Poisson timing model.

17. The apparatus of claim 13 wherein the model of time-dependent behavior comprises a Poisson timing model having deterministic shifts in one or more transaction rates.

18. The apparatus of claim 13 wherein the event-driven estimator for a given one of the periods comprises a weighted average of a previous reciprocal transaction rate and a time since a previous transaction in the given period, and wherein the event-driven estimator for at least a subset of the other periods increases by an amount of time that has been spent in the given period since the previous transaction.

19. The apparatus of claim 13 wherein the event-driven estimator is updated for every transaction in a specified one of the plurality of periods and at the end of each of at least a subset of the plurality of periods.

20. The apparatus of claim 13 wherein the event-driven estimator is updated for every transaction, regardless of which of the plurality of periods any particular transaction falls in, but not at the end of each of the plurality of periods.

21. The apparatus of claim 13 wherein the event-driven estimator is updated for a given one of the plurality of periods only when there is a transaction occurring within that period.

22. The apparatus of claim 13 wherein the event-driven estimator is updated once for a given one of the plurality of periods after a plurality of transactions occurring within that period.

23. The apparatus of claim 13 wherein the event-driven estimator first generates an estimated transaction rate {circumflex over (λ

)}_j,n for period j and a given transaction n, and then generates an estimated period probability {circumflex over (π

)}_j,n for period j as {circumflex over (λ

)}_j,n/Σ

_k{circumflex over (λ

)}_k,n.

24. The apparatus of claim 13 wherein an estimated transaction rate {circumflex over (λ

)}_j,n under the model satisfies ${\widehat{\mathrm{\lambda <br><br>}}}_{j,n}^{-1}=\{\begin{array}{cc}\left(1-w\right)<br><br>{\widehat{\mathrm{\lambda <br><br>}}}_{j,n-1}^{-1}+{\mathrm{wZ}}_{j,n,}& \mathrm{if}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{transaction}<br><br>\text{\hspace{1em}<br><br>}<br><br>n<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{falls}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{in}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{period}<br><br>\text{\hspace{1em}<br><br>}<br><br>j\\ {\widehat{\mathrm{\lambda <br><br>}}}_{j,n-1}^{-1}+\frac{w}{1-w}<br><br>Z_{j,n,}& \mathrm{if}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{transaction}<br><br>\text{\hspace{1em}<br><br>}<br><br>n<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{does}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{not}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{fall}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{in}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{period}<br><br>\text{\hspace{1em}<br><br>}<br><br>j,\end{array}$

wherein w denotes a specified updating weight, Z_j,n is an elapsed time in period j since a previous transaction n−

1, and {circumflex over (λ

)}_j,n<sup>−

1</sup> is a reciprocal rate that estimates the average time between transactions in period j.

25. An article of manufacture comprising a machine-readable storage medium for storing one or more programs for use in updating a record in a database system in conjunction with one or more transactions involving at least a portion of the record, wherein the one or more programs when executed implement the steps of:

determining a record affected by a given one of the transactions;

retrieving a current version of the record from a memory of the system;

estimating at least one of a transaction rate and a period probability for the record using an event-driven estimator based at least in part on application of at least a portion of a stochastic model of time-dependent behavior to the current version of the record, the model having associated therewith a plurality of periods and a number of period-based transaction rates; and

generating an updated version of the record based on the event-driven estimator.