Method and system using association rules to form custom lists of cookies
View Patent ↗A targeting rule set of association rules may be created. A plurality of association rules may be selected. A plurality of candidate rule sets may be constructed based on the plurality of association rules. The plurality of candidate rule sets may be evaluated to produce metrics. One of the plurality of candidate rule sets may be assigned as the targeting rule set based on the metrics. A custom list of user may be formed using association rules. A user may be determined to belong to a segment by applying one or more rules of a targeting rule set to user attribute data. A custom list of cookies to show advertising may be formed using combinations of association rules.
1. A method for creating a targeting rule set of association rules, the method comprising:
selecting, with a computer, a plurality of association rules;
determining, with the computer, complexity parameters, a performance parameter, and a stopping parameter;
selecting, with the computer, candidate rules from the plurality of association rules;
calculating, with the computer, quality values of the candidate rules;
removing, with the computer, an underperforming rule that does not meet the performance parameter from the candidate rules;
iterating, with the computer, through the complexity parameters to generate a plurality of candidate rule sets until the stopping parameter is satisfied, wherein the complexity parameters correspond to a plurality of working rules of the candidate rules;
evaluating, with the computer, the plurality of candidate rule sets to produce a metric; and
assigning, with the computer, one of the plurality of candidate rule sets as the targeting rule set based on the metric.
2. The method of claim 1 , wherein iterating through the complexity parameters to generate the plurality of candidate rule sets further includes:
determining a first remaining rule from the plurality of working rules based on the quality values of the first remaining rule;
adding the first remaining rule to the candidate rule set;
removing the first remaining rule from the plurality of working rules;
removing a transaction from a scale dataset and from a training dataset, wherein the transaction meets a support of the first remaining rule; and
when the stopping parameter is not satisfied, recalculating conditional quality values for the scale dataset and the training dataset.
3. The method of claim 2 , wherein the quality values include at least one of:
a support of a rule from the plurality of working rules with respect to the scale dataset,
a confidence of the rule from the plurality of working rules with respect to the scale dataset,
a coverage of the rule from the plurality of working rules with respect to the scale dataset,
a scale of the rule from the plurality of working rules with respect to the scale dataset,
a lift of the rule from the plurality of working rules with respect to the scale dataset, and any combination thereof.
4. The method of claim 1 , wherein the complexity parameters include a minimum per-rule support threshold.
5. A system for creating a targeting rule set of association rules, the system comprising:
a memory; and
a processor configured to:
select a plurality of association rules;
determine complexity parameters, a performance parameter, and a stopping parameter;
select candidate rules from the plurality of association rules;
calculate quality values of the candidate rules;
remove an underperforming rule that does not meet the performance parameter from the candidate rules;
iterate through the complexity parameters to generate a plurality of candidate rule sets until the stopping parameter is satisfied, wherein the complexity parameters correspond to a plurality of working rules of the candidate rules;
evaluate the plurality of candidate rule sets to produce a metric; and
assign one of the plurality of candidate rule sets as the targeting rule set based on the metric.
6. The system of claim 5 , wherein the processor being configured to iterate through the complexity parameters to generate the plurality of candidate rule sets further includes the processor being configured to:
determine a first remaining rule from the plurality of working rules based on the quality values of the first remaining rule;
add the first remaining rule to the candidate rule set;
remove the first remaining rule from the plurality of working rules;
remove a transaction from a scale dataset and from a training dataset, wherein the transaction meets a support of the first remaining rule; and
when the stopping parameter is not satisfied, recalculate conditional quality values for the scale dataset and the training dataset.
7. The system of claim 6 , wherein the quality values include at least one of:
a support of a rule from the plurality of working rules with respect to the scale dataset,
a confidence of the rule from the plurality of working rules with respect to the scale dataset,
a coverage of the rule from the plurality of working rules with respect to the scale dataset,
a scale of the rule from the plurality of working rules with respect to the scale dataset,
a lift of the rule from the plurality of working rules with respect to the scale dataset, and any combination thereof.
8. The system of claim 7 , wherein the complexity parameters include a minimum per-rule support threshold.
9. A computer readable storage medium for creating a targeting rule set of association rules, the computer readable storage medium comprising instructions that if executed enable a computing system to:
select a plurality of association rules;
determine complexity parameters, a performance parameter, and a stopping parameter;
select candidate rules from the plurality of association rules;
calculate quality values of the candidate rules;
remove an underperforming rule that does not meet the performance parameter from the candidate rules;
iterate through the complexity parameters to generate a plurality of candidate rule sets until the stopping parameter is satisfied, wherein the complexity parameters correspond to a plurality of working rules of the candidate rules;
evaluate the plurality of candidate rule sets to produce a metric; and
assign one of the plurality of candidate rule sets as the targeting rule set based on the metric.
10. The computer readable storage medium of claim 9 , wherein the instructions for iterating through the complexity parameters to generate the plurality of candidate rule sets further includes instructions that if executed enable the computing system to:
determine a first remaining rule from the plurality of working rules based on the quality values of the first remaining rule;
add the first remaining rule to the candidate rule set;
remove the first remaining rule from the plurality of working rules;
remove a transaction from a scale dataset and from a training dataset, wherein the transaction meets a support of the first remaining rule; and
when the stopping parameter is not satisfied, recalculate conditional quality values for the scale dataset and the training dataset.
11. The computer readable storage medium of claim 10 , wherein the quality values include at least one of:
a support of a rule from the plurality of working rules with respect to the scale dataset,
a confidence of the rule from the plurality of working rules with respect to the scale dataset,
a coverage of the rule from the plurality of working rules with respect to the scale dataset,
a scale of the rule from the plurality of working rules with respect to the scale dataset,
a lift of the rule from the plurality of working rules with respect to the scale dataset, and any combination thereof.
12. The computer readable storage medium of claim 9 , wherein the complexity parameters include a minimum per-rule support threshold.
13. A method for using combinations of association rules to form a custom list of cookies to show advertising, the method comprising:
receiving first party data from a provider via files, tags, or any combination thereof;
constructing association rules using an association algorithm;
applying the association rules to an unlabeled data set to calculate a projected scale of each rule;
(a) filtering rules that have a scale above an initial threshold into a remaining rule set;
(b) choosing a rule having a highest lift in the remaining rule set;
(c) removing users that fall under the rule from a training set and from the unlabeled data set;
(d) recalculating a lift value, a confidence value, a scale value, a support value, and a coverage value for rules in the remaining rule set;
(e) repeating (b)-(d) until a goal projected segment size has been reached;
repeating (a)-(e) while varying the initial threshold to choose an optimal threshold size, wherein the rule sets are validated using a validation set f-measure;
selecting a subset of the association rules based on the application of the association rules to the unlabeled data set;
placing the subset of the association rules in a platform;
creating a segment using the platform; and
selling the segment to the provider of the first party data.
14. A system for using combinations of association rules to form a custom list of cookies to show advertising, the system comprising:
a memory; and
a processor configured to:
receive first party data from a provider via files, tags, or any combination thereof;
construct association rules using an association algorithm;
apply the association rules to an unlabeled data set to calculate a projected scale of each rule;
(a) filter rules that have a scale above an initial threshold into a remaining rule set;
(b) choose a rule having a highest lift in the remaining rule set;
(c) remove users that fall under the rule from a training set and from the unlabeled data set;
(d) recalculate a lift value, a confidence value, a scale value, a support value, and a coverage value for rules in the remaining rule set;
(e) repeat (b)-(d) until a goal projected segment size has been reached;
repeat (a)-(e) while varying the initial threshold to choose an optimal threshold size, wherein the rule sets are validated using a validation set f-measure;
select a subset of the association rules based on the application of the association rules to the unlabeled data set;
place the subset of the association rules in a platform;
create a segment using the platform; and
sell the segment to the provider of the first party data.
15. A computer readable medium for using combinations of association rules to form a custom list of cookies to show advertising, the computer readable medium comprising instructions that if executed enable a computing system to:
receive first party data from a provider via files, tags, or any combination thereof;
construct association rules using an association algorithm;
apply the association rules to an unlabeled data set to calculate a projected scale of each rule;
(a) filter rules that have a scale above an initial threshold into a remaining rule set;
(b) choose a rule having a highest lift in the remaining rule set;
(c) remove users that fall under the rule from a training set and from the unlabeled data set;
(d) recalculate a lift value, a confidence value, a scale value, a support value, and a coverage value for rules in the remaining rule set;
(e) repeat (b)-(d) until a goal projected segment size has been reached;
repeat (a)-(e) while varying the initial threshold to choose an optimal threshold size, wherein the rule sets are validated using a validation set f-measure;
select a subset of the association rules based on the application of the association rules to the unlabeled data set;
place the subset of the association rules in a platform;
create a segment using the platform; and
sell the segment to the provider of the first party data.