Push-based recommendations

摘要

Among other things, one or more techniques and/or systems are provided for pushing a recommendation to a user. That is, a recommendation may be pushed to a device of the user based upon a triggering event associated with the user. The recommendation may be provided, for example, without user solicitation for the recommendation. In one example, a recommendation component may observe that the user frequently stops for ice cream on Fridays after work (e.g., based upon prior social network check-ins). Accordingly, on the following Friday, the recommendation component may push a recommendation to the user's device to visit a particular grocery store on the way home from work that is within 10 minutes of the user's home so that the user can avoid melting ice cream (e.g., a location constraint may be applied to choose a grocery store that is relatively close to the user's home).

主权项

1. A method for pushing a recommendation to a user, comprising:
identifying a triggering event associated with a user; determining one or more suggested entities corresponding to the triggering event, the one or more suggested entities comprising a first suggested entity and a second suggested entity, the first suggested entity having a first rank in association with a first factor, the second suggested entity having a second rank in association with the first factor, the first suggested entity having a third rank in association with a second factor, the second suggested entity having a fourth rank in association with the second factor, the first rank lower than the second rank, the third rank higher than the fourth rank; constructing an entity selection criteria model based upon one or more constraints derived from observation data associated with the triggering event; applying the entity selection criteria model to the one or more suggested entities to create a recommendation for the triggering event, the recommendation comprising the first suggested entity but not the second suggested entity based upon the first rank being lower than the second rank and the third rank being higher than the fourth rank; and pushing the recommendation to a device associated with the user.