Semantic based collaborative filtering
First Claim
1. A method for providing product recommendations to customers in an e-commerce environment, comprising the steps of:
- extracting content information from products corresponding to a plurality of customers;
generating content representations of the products from the content information;
generating compatibility representations of the products;
calculating a similarity function between pairs of content attributes corresponding to the products;
calculating a similarity function between pairs of compatibility attributes corresponding to the products;
clustering the plurality of customers into a plurality of peer groups;
for a given customer, determining a closest peer group of the plurality of peer groups based on the similarity functions; and
generating at least one potential recommendation for the given customer based on the closest peer group and an absence of explicit customer ratings.
2 Assignments
0 Petitions
Accused Products
Abstract
A method for providing product recommendations to customers in an e-commerce environment includes the step of generating content and compatibility representations of products corresponding to a plurality of customers. A similarity function is calculated between pairs of content attributes corresponding to the products. A similarity function is calculated between pairs of compatibility attributes corresponding to the products. The plurality of customers are clustered into a plurality of peer groups. For a given customer, a closest peer group of the plurality of peer groups is determined. At least one potential recommendation is then generated for the given customer based on the closest peer group.
175 Citations
26 Claims
-
1. A method for providing product recommendations to customers in an e-commerce environment, comprising the steps of:
-
extracting content information from products corresponding to a plurality of customers;
generating content representations of the products from the content information;
generating compatibility representations of the products;
calculating a similarity function between pairs of content attributes corresponding to the products;
calculating a similarity function between pairs of compatibility attributes corresponding to the products;
clustering the plurality of customers into a plurality of peer groups;
for a given customer, determining a closest peer group of the plurality of peer groups based on the similarity functions; and
generating at least one potential recommendation for the given customer based on the closest peer group and an absence of explicit customer ratings. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13)
concatenating product descriptions of the products bought by an individual customer; and
calculating a content vector from the concatenated product descriptions.
-
-
3. The method according to claim 1, wherein said step of generating the content representations further comprises the steps of extracting the product descriptions, when the individual customer at least one of browses and purchases the products.
-
4. The method according to claim 1, wherein said step of generating the compatibility representations further comprises the steps of:
-
calculating a fraction of time that a categorical attribute takes on a given value; and
calculating a compatibility vector from the calculated fraction of time.
-
-
5. The method according to claim 1, wherein said clustering step comprises the step of calculating a similarity function between a pair of categorical values as a predefined function of a support of the pair of categorical values.
-
6. The method according to claim 1, wherein said clustering step comprises the steps of:
-
calculating a content-similarity between a content vector of a particular customer and content centroids of the plurality of customers; and
calculating a compatibility-similarity between a compatibility vector of the particular customer and compatibility centroids of the plurality of customers.
-
-
7. The method according to claim 1, wherein said step of determining the closest peer group comprises the steps of:
-
determining a predetermined number of closest clusters to the given customer; and
designating a union of the closest clusters as the closest peer group.
-
-
8. The method according to claim 1, wherein said step of generating the at least one potential recommendation comprises the steps of:
-
determining most frequently bought products by the customers in the closest peer group; and
recommending at least one of the most frequently bought products to the given customer.
-
-
9. The method according to claim 1, further comprising the step of filtering the at least one potential recommendation using domain specific rules.
-
10. The method according to claim 9, wherein said filtering step comprises the steps of:
-
determining whether any of the domain specific rules are relevant to the given customer and the at least one potential recommendation;
determining whether any of the relevant domain specific rules are violated, when any of the domain specific rules are relevant to the given customer and the at least one potential recommendation; and
providing the at least one potential recommendation to the given customer, when any of the relevant domain specific rules are not violated.
-
-
11. The method according to claim 10, wherein said filtering step further comprises the step of providing the at least one potential recommendation to the given customer, when none of the domain specific rules are relevant to the given customer and the at least one potential recommendation.
-
12. The method according to claim 10, wherein a domain specific rule is relevant for the given customer and the at least one potential recommendation when antecedent conditions of the domain specific rule match a buying history of the given customer and a result of the rule satisfies the at least one potential recommendation.
-
13. The method according to claim 10, wherein a domain specific rule is violated when the domain specific rules is relevant to the given customer and the at least one potential recommendation, but a key compatibility attribute does not match the at least one potential recommendation.
-
14. A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform steps for providing product recommendations to customers in an e-commerce environment, said method steps comprising:
-
generating content and compatibility representations of products corresponding to a plurality of customers;
calculating a similarity function between pairs of content attributes corresponding to the products;
calculating a similarity function between pairs of compatibility attributes corresponding to the products;
clustering the plurality of customers into a plurality of peer groups;
for a given customer, determining a closest peer group of the plurality of peer groups based on the similarity functions; and
generating at least one potential recommendation for the given customer based on the closest peer group and an absence of explicit customer ratings. - View Dependent Claims (15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26)
concatenating product descriptions of the products bought by an individual customer; and
calculating a content vector from the concatenated product descriptions.
-
-
16. The program storage device according to claim 14, wherein said step of generating the content representations further comprises the steps of extracting the product descriptions, when the individual customer at least one of browses and purchases the products.
-
17. The program storage device according to claim 14, wherein said step of generating the compatibility representations further comprises the steps of:
-
calculating a fraction of time that a categorical attribute takes on a given value; and
calculating a compatibility vector from the calculated fraction of time.
-
-
18. The program storage device according to claim 14, wherein said clustering step comprises the step of calculating a similarity function between a pair of categorical values as a predefined function of a support of the pair of categorical values.
-
19. The program storage device according to claim 14, wherein said clustering step comprises the steps of:
-
calculating a content-similarity between a content vector of a particular customer and content centroids of the plurality of customers; and
calculating a compatibility-similarity between a compatibility vector of the particular customer and compatibility centroids of the plurality of customers.
-
-
20. The program storage device according to claim 14, wherein said step of determining the closest peer group comprises the steps of:
-
determining a predetermined number of closest clusters to the given customer; and
designating a union of the closest clusters as the closest peer group.
-
-
21. The program storage device according to claim 14, wherein said step of generating the at least one potential recommendation comprises the steps of:
-
determining most frequently bought products by the customers in the closest peer group; and
recommending at least one of the most frequently bought products to the given customer.
-
-
22. The program storage device according to claim 14, further comprising the step of filtering the at least one potential recommendation using domain specific rules.
-
23. The program storage device according to claim 22, wherein said filtering step comprises the steps of:
-
determining whether any of the domain specific rules are relevant to the given customer and the at least one potential recommendation;
determining whether any of the relevant domain specific rules are violated, when any of the domain specific rules are relevant to the given customer and the at least one potential recommendation; and
providing the at least one potential recommendation to the given customer, when any of the relevant domain specific rules are not violated.
-
-
24. The program storage device according to claim 23, wherein said filtering step further comprises the step of providing the at least one potential recommendation to the given customer, when none of the domain specific rules are relevant to the given customer and the at least one potential recommendation.
-
25. The program storage device according to claim 23, wherein a domain specific rule is relevant for the given customer and the at least one potential recommendation when antecedent conditions of the domain specific rule match a buying history of the given customer and a result of the rule satisfies the at least one potential recommendation.
-
26. The program storage device according to claim 23, wherein a domain specific rule is violated when the domain specific rules is relevant to the given customer and the at least one potential recommendation, but a key compatibility attribute does not match the at least one potential recommendation.
Specification