Method and medium for determining whether a target item is related to a candidate affinity item
First Claim
Patent Images
1. A method for recommending affinity products, the method comprising:
- collecting data corresponding to monitored actions of a user of a web site, the monitored actions collected over a plurality of browsing sessions;
aggregating said collected data into aggregate groups, wherein said collected data comprises speedometer data which is aggregated and stored for values exceeding a threshold value in order to limit RAM usage;
identifying, with a server computer, at least one affinity product related to a target product, the identifying based on the monitored actions and a weighted formula, wherein weights are used to specify different probabilities to interrelate actions performed by said user across sessions while visiting said web site, wherein said weights combine a common affiliated product in a buy-to-buy, view-to-view, view-to-buy and abandon-to-buy actions;
receiving a selection of variables a, b, c and d assigned a percentage between 0 and 100% to select a recommended affinity product according to said weighted formula;
(a×
BB)+(b×
VV)+(c×
VB)+(d×
AB),wherein said BB corresponds to said buy-to-buy action, wherein said VV corresponds to said view-to-view action, wherein said VB corresponds to said view-to-buy action, and wherein said AB corresponds to said abandon-to-buy action;
displaying, with the server computer, a preview of the at least one affinity product and said stored speedometer data shown in a report module of a speedometer;
modifying at least one weight of the weighted formula; and
displaying, with the server computer, an updated preview showing any change in the at least one affinity product due to the modifying.
5 Assignments
0 Petitions
Accused Products
Abstract
Recommendations for purchase are made based on customer behavior across multiple sessions. Correlations used for recommendations include: buy-to-buy (cross-session), view-to-view (same-session), view-to-buy (same-session), and abandon-to-buy (same-session) actions. A preview display allows a merchant to adjust recommendation algorithm weightings to achieve a desired result. A closed-loop system is provided with real-time feedback. The recommendations can be based on various segments of other users, including users of the same search engine.
14 Citations
20 Claims
-
1. A method for recommending affinity products, the method comprising:
-
collecting data corresponding to monitored actions of a user of a web site, the monitored actions collected over a plurality of browsing sessions; aggregating said collected data into aggregate groups, wherein said collected data comprises speedometer data which is aggregated and stored for values exceeding a threshold value in order to limit RAM usage; identifying, with a server computer, at least one affinity product related to a target product, the identifying based on the monitored actions and a weighted formula, wherein weights are used to specify different probabilities to interrelate actions performed by said user across sessions while visiting said web site, wherein said weights combine a common affiliated product in a buy-to-buy, view-to-view, view-to-buy and abandon-to-buy actions; receiving a selection of variables a, b, c and d assigned a percentage between 0 and 100% to select a recommended affinity product according to said weighted formula;
(a×
BB)+(b×
VV)+(c×
VB)+(d×
AB),wherein said BB corresponds to said buy-to-buy action, wherein said VV corresponds to said view-to-view action, wherein said VB corresponds to said view-to-buy action, and wherein said AB corresponds to said abandon-to-buy action; displaying, with the server computer, a preview of the at least one affinity product and said stored speedometer data shown in a report module of a speedometer; modifying at least one weight of the weighted formula; and displaying, with the server computer, an updated preview showing any change in the at least one affinity product due to the modifying. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8)
-
-
9. A method for recommending affinity products, the method comprising:
-
collecting data corresponding to monitored actions of a user of a web site, the monitored actions collected over a plurality of browsing sessions; aggregating said collected data into aggregate groups, wherein said collected data comprises speedometer data which is aggregated and stored for values exceeding a threshold value in order to limit RAM usage; providing a formula for identifying affinity products based on said monitored actions, wherein weights are used to specify different probabilities to interrelate actions performed by said user across sessions while visiting said web site, wherein said weights combine a common affiliated product in a buy-to-buy, view-to-view, view-to-buy and abandon-to-buy actions; receiving a selection of variables a, b, c and d assigned a percentage between 0 and 100% to select a recommended affinity product according to said formula;
(a×
BB)+(b×
VV)+(c×
VB)+(d×
AB),wherein said BB corresponds to said buy-to-buy action, wherein said VV corresponds to said view-to-view action, wherein said VB corresponds to said view-to-buy action, and wherein said AB corresponds to said abandon-to-buy action; tracking said monitored actions for an identified segment of users; selecting, with a server computer, at least one affinity product using said formula and said monitored actions of said segment of users; and displaying, with the server computer, a preview of the at least one affinity product and said stored speedometer data shown in a report module of a speedometer. - View Dependent Claims (10, 11, 12, 13, 14, 15, 16)
-
-
17. A computer program product for recommending affinity products, the computer program product comprising a non-transitory computer readable storage medium having program code embodied therewith, the program code comprising the programming instructions for:
-
collecting data corresponding to monitored actions of a user of a web site, the monitored actions collected over a plurality of browsing sessions; aggregating said collected data into aggregate groups, wherein said collected data comprises speedometer data which is aggregated and stored for values exceeding a threshold value in order to limit RAM usage; identifying, with a server computer, at least one affinity product related to a target product, the identifying based on the monitored actions and a weighted formula, wherein weights are used to specify different probabilities to interrelate actions performed by said user across sessions while visiting said web site, wherein said weights combine a common affiliated product in a buy-to-buy, view-to-view, view-to-buy and abandon-to-buy actions; receiving a selection of variables a, b, c and d assigned a percentage between 0 and 100% to select a recommended affinity product according to said weighted formula;
(a×
BB)+(b×
VV)+(c×
VB)+(d×
AB),wherein said BB corresponds to the buy-to-buy action, wherein said VV corresponds to the view-to-view action, wherein said VB corresponds to the view-to-buy action, and wherein said AB corresponds to the abandon-to-buy action; displaying, with the server computer, a preview of the at least one affinity product and said stored speedometer data shown in a report module of a speedometer; modifying at least one weight of the weighted formula; and displaying, with the server computer, an updated preview showing any change in the at least one affinity product due to the modifying. - View Dependent Claims (18, 19, 20)
-
Specification