Serendipity generating method, system, and device

摘要

A serendipity generating method, system, and device generates recommendations by identifying contrasting corresponding topic affinity values between users who more generally have a relatively high degree of similarity between corresponding topic affinity values. The topic affinity values may represent user interests with respect to topics and the topic affinity values may be inferred from usage behaviors, including geographic location information. Recommendations may be further in accordance with an assessment of the amount or quality of the usage behaviors from which the topic affinity values are derived and/or in accordance with the application of a probabilistic process. The recommendations may comprise computer-implemented objects that have relatively high affinities to topics associated with the contrasting corresponding topic affinity values.

主权项

1. A computer-implemented method, comprising:
selecting automatically a first affinity vector comprising a first plurality of topic affinity level values that are associated with a first user of a computer-implemented system, wherein a plurality of the first plurality of topic affinity level values are each based on an inference from a first one or more usage behaviors; selecting automatically a second affinity vector comprising a second plurality of topic affinity level values that are associated with a second user of the computer-implemented system, wherein a plurality of the second plurality of topic affinity level values are each based on an inference from a second one or more usage behaviors, and wherein the selection of the second affinity vector is in accordance with a determination of a relatively high level of similarity between a plurality of the first plurality of topic affinity level values and a corresponding plurality of the second plurality of topic affinity level values; identifying automatically one or more pairs of contrasting corresponding topic affinity level values by comparing topic affinity level values in the first affinity vector with topic affinity level values in the second affinity vector; and generating automatically a recommendation for delivery to the first user, wherein the recommendation is generated in accordance with the one or more pairs of contrasting corresponding topic affinity level values.