| CPC G06F 16/9535 (2019.01) [G06F 16/24578 (2019.01); G06F 16/36 (2019.01)] | 20 Claims |

|
1. A computer-implemented method of refining an output of an information retrieval system based on user preferences of a user associated with a data consumer, wherein the information retrieval system is configured to facilitate access to information items of digital content, the information items comprising information on a plurality of items, the information including a plurality of item attributes of the plurality of items, wherein a respective item of the plurality of items is any suitable object or information about which the user may express preferences, and wherein a respective information item comprises item attributes corresponding to the respective item, the method comprising:
using at least one hardware processor to perform:
deriving, by a preference engine, from the user preferences, at least one first-order preference, wherein the at least one first-order preference comprises a preference among item attributes of at least one of the plurality of items;
encoding, by the preference engine, a first-order preference model comprising first-order preference attributes, wherein the first-order preference attributes comprise the user first-order preferences encoded as attributes based on at least a subset of the at least one first attribute of the first-order preferences;
deriving, by the preference engine at least one second-order preference associated with the first-order preference, wherein the at least one second-order preference comprises a preference among item attributes of the plurality of items;
encoding, by the preference engine, a second-order preference model comprising second-order preference attributes, wherein the second-order preference attributes comprise the user second-order preferences encoded as attributes based on the plurality of second attributes of the second-order preferences;
encoding a preference graph from the first-order preference model and the second-order preference model; and
accessing, by the information retrieval system, the preference graph to rank, based on the first-order and second-order preference attributes, at least a subset of the plurality of information items to obtain a ranking of the information items in the subset.
|