Concerted learning and multi-instance sequential prediction tree

Patent metadata

Field	Value
Publication number	US-11017452-B2
Application number	US-201816154903-A
Country	US
Kind code	B2
Filing date	Oct 9, 2018
Priority date	Oct 9, 2018
Publication date	May 25, 2021
Grant date	May 25, 2021

Abstract

A method, system and computer readable medium for performing a purchase prediction operation. The purchase prediction operation includes: selecting a target purchaser, the purchase prediction operation providing a purchase prediction for the target purchaser; capturing a product term associated with a most recent purchase period of the target purchaser; performing a sequential recommendation operation, the sequential recommendation operation providing a sequence recommendation score; and, generating a purchase pattern prediction for the target user based upon the sequential recommendation score.

First claim

What is claimed is:

1. A computer-implementable method for performing a purchase prediction operation, comprising:
selecting a target purchaser, the purchase prediction operation providing a purchase prediction for the target purchaser;
capturing a product term associated with a most recent purchase period of the target purchaser;
performing a sequential recommendation operation, the sequential recommendation operation providing a sequential recommendation score, the sequential recommendation score being based upon a number of historical purchase periods, a number of future purchase periods, and a similarity quotient score, the similarity quotient score representing a degree of similarity between historical purchase behavior of the target purchaser and historical purchase behavior of at least one neighbor purchaser, the-neighbor purchaser comprising a purchaser in a set of purchasers, the set of purchasers including the target purchaser and another neighbor purchaser;
using the similarity quotient score to select neighbor purchasers with similar historical purchase behavior; and,
generating a purchase pattern prediction for the target purchaser based upon the sequential recommendation score and the neighbor purchasers with similar historical purchase behavior.

2. The method of claim 1 , wherein:
the sequential recommendation operation comprises a concerted learning component (CLC) and a multi-instance sequential component (MISC).

3. The method of claim 1 , wherein:
the sequential recommendation operation uses historical purchase patterns associated with similar customers to provide predictions of future purchases for the purchaser.

4. The method of claim 1 , wherein:
the sequential recommendation operation is implemented to search for most recent list of products (MRLP) purchased by a target customer during a most recent purchase period.

5. The method of claim 4 , wherein:
searching for the MRLP is performed for similar customers, the result of searching for the MRLP for similar customers being used to generate a prediction tree, the prediction tree being based on a multi-instance occurrence of MRLP.

6. The method of claim 5 , wherein:
the prediction tree is implemented to capture multiple sets of ‘n’ purchase sequences (n-PS) after the MRLP; and,
MISC scores are computed for each purchase sequence n-PS.

7. A system comprising:
a processor;
a data bus coupled to the processor; and
a non-transitory, computer-readable storage medium embodying computer program code, the non-transitory, computer-readable storage medium being coupled to the data bus, the computer program code interacting with a plurality of computer operations and comprising instructions executable by the processor and configured for:
selecting a target purchaser, the purchase prediction operation providing a purchase prediction for the target purchaser;
capturing a product term associated with a most recent purchase period of the target purchaser;
performing a sequential recommendation operation, the sequential recommendation operation providing a sequential recommendation score, the sequential recommendation score being based upon a number of historical purchase periods, a number of future purchase periods, and a similarity quotient score, the similarity quotient score representing a degree of similarity between historical purchase behavior of the target purchaser and historical purchase behavior of at least one neighbor purchaser, the-neighbor purchaser comprising a purchaser in a set of purchasers, the set of purchasers including the target purchaser and another neighbor purchaser;
using the similarity quotient score to select neighbor purchasers with similar historical purchase behavior; and,
generating a purchase pattern prediction for the target purchaser based upon the sequential recommendation score and the neighbor purchasers with similar historical purchase behavior.

8. The system of claim 7 , wherein:
the sequential recommendation operation comprises a concerted learning component (CLC) and a multi-instance sequential component (MISC).

9. The system of claim 7 , wherein:
the sequential recommendation operation uses historical purchase patterns associated with similar customers to provide predictions of future purchases for the purchaser.

10. The system of claim 9 , wherein:
the sequential recommendation operation is implemented to search for most recent list of products (MRLP) purchased by a target customer during a most recent purchase period.

11. The system of claim 10 , wherein the instructions executable by the processor are further configured for:
searching for the MRLP is performed for similar customers, the result of searching for the MRLP for similar customers being used to generate a prediction tree, the prediction tree being based on a multi-instance occurrence of MRLP.

12. The system of claim 11 , wherein:
the prediction tree is implemented to capture multiple sets of ‘n’ purchase sequences (n-PS) after the MRLP; and,
MISC scores are computed for each purchase sequence n-PS.

13. A non-transitory, computer-readable storage medium embodying computer program code, the computer program code comprising computer executable instructions configured for:
selecting a target purchaser, the purchase prediction operation providing a purchase prediction for the target purchaser;
capturing a product term associated with a most recent purchase period of the target purchaser;
performing a sequential recommendation operation, the sequential recommendation operation providing a sequential recommendation score, the sequential recommendation score being based upon a number of historical purchase periods, a number of future purchase periods, and a similarity quotient score, the similarity quotient score representing a degree of similarity between historical purchase behavior of the target purchaser and historical purchase behavior of at least one neighbor purchaser, the-neighbor purchaser comprising a purchaser in a set of purchasers, the set of purchasers including the target purchaser and another neighbor purchaser;
using the similarity quotient score to select neighbor purchasers with similar historical purchase behavior; and,
generating a purchase pattern prediction for the target purchaser based upon the sequential recommendation score and the neighbor purchasers with similar historical purchase behavior.

14. The non-transitory, computer-readable storage medium of claim 13 , wherein:
the sequential recommendation operation comprises a concerted learning component (CLC) and a multi-instance sequential component (MISC).

15. The non-transitory, computer-readable storage medium of claim 13 , wherein:
the sequential recommendation operation uses historical purchase patterns associated with similar customers to provide predictions of future purchases for the purchaser.

16. The non-transitory, computer-readable storage medium of claim 15 , wherein:
the sequential recommendation operation is implemented to search for most recent list of products (MRLP) purchased by a target customer during a most recent purchase period.

17. The non-transitory, computer-readable storage medium of claim 16 , wherein the computer executable instructions are further configured for:
searching for the MRLP is performed for similar customers, the result of searching for the MRLP for similar customers being used to generate a prediction tree, the prediction tree being based on a multi-instance occurrence of MRLP.

18. The non-transitory, computer-readable storage mediu

Assignees

• Dell Products Lp

Inventors

• Kanagovi Ramakanth
• Chowdhury Arnab
• Sahoo Sumant

Classifications

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06N20/00

  Machine learning · CPC title

• G06Q30/0631 Primary

  Recommending goods or services · CPC title

• G06N5/02

  Knowledge representation; Symbolic representation · CPC title