System and method for the dynamic generation of correlation scores between arbitrary objects
View Patent ↗Systems and methods are described for performing the dynamic generation of correlation scores between arbitrary objects. When a behavioral event is recorded, that is to say when an end user interacts with multiple objects, relationships between objects are created. These relationships are maintained as a list. When a request for correlated items is requested based upon a seed object, a list of correlated items is dynamically created through the generation of a pivot set and a scoring algorithm to compute the list of correlated items.
1. A network comprising:
a plurality of content servers;
a client computer to access and establish relationships with one or more of the content servers; and
a management server to receive relationship information from the servers and to perform collaborative filtering on the relationship information to generate recommendations for the client, the collaborative filtering including:
generating a seed object;
retrieving a list of objects corresponding to the seed object from a storage table;
restricting the list of objects based upon a parameter list to generate a pivot set of objects;
generating a candidate set by determining objects that have interacted with each of the pivot set of objects;
computing a score for each member of the candidate set; and
generating a list of correlated items based upon at least one score.
2. The network of claim 1 , wherein the management server further comprises:
a communications module configured to communicate with one or more of the content servers and the client computer; and
a normalization module configured to normalize messages from one or more of the content servers and the client computer.
3. The network of claim 2 , wherein the normalization module analyzes predicates within the received messages and verifies that the labels associated within the predicates are in expected formats.
4. The network of claim 3 , wherein the normalization module modifies the labels by adding or removing details if the labels are not in an expected format.
5. The network of claim 4 , wherein the management server further comprises:
a canonicalization module to canonicalize the normalized messages by associating labels, predicates, and intents with unique fixed-width integer values; and
a storage module to store a table of canonicalized relationships.
6. The network of claim 5 , wherein the canonicalization module receives and extracts the labels, predicates and optional intents from the normalized messages and looks up a corresponding integer value in the table.
7. The network of claim 6 , wherein the management server further comprises a recommendation generation module to generate recommendations for the client computer based on the relationships stored in the storage module.
8. The article of manufacture of claim 1 , wherein the score is calculated using a Pearson's R algorithm.
9. A management server comprising:
a storage module to store a table of relationship information between one or more content servers and a client computer; and
a recommendation generation module to perform collaborative filtering on the stored relationship information and to generate recommendations for a client computer, the collaborative filtering including:
generating a seed object;
retrieving a list of objects corresponding to the seed object from a storage table;
restricting the list of objects based upon a parameter list to generate a pivot set of objects;
generating a candidate set by determining objects that have interacted with each of the pivot set of objects;
computing a score for each member of the candidate set; and
generating a list of correlated items based upon at least one score.
10. The management server of claim 9 , wherein the management server further comprises:
a communications module configured to communicate with one or more of the content servers and the client computer; and
a normalization module configured to normalize messages from one or more of the content servers and the client computer.
11. The management server of claim 10 , wherein the normalization module analyzes predicates within the received messages and verifies that the labels associated within the predicates are in expected formats.
12. The management server of claim 11 , wherein the normalization module modifies the labels by adding or removing details if the labels are not in an expected format.
13. The management server of claim 12 , wherein the management server further comprises:
a canonicalization module to canicalize the normalized messages by associating labels, predicates, and intents with unique fixed-width integer values.
14. The management server of claim 13 , wherein the canonicalization module receives and extracts the labels, predicates and optional intents from the normalized messages and looks up an corresponding integer value in the table.