Context-based organization of digital media

One or more non-transitory computer storage media storing computer-useable instructions that, when used by one or more computing devices, cause the one or more computing devices to perform operations comprising:

• receiving a set of images having a plurality of associated content tags;
• determining, based on matching between the plurality of associated content tags and a set of learned content tags representative a category of event types, and based on weighting a learned correlation score that quantifies a likelihood that a learned content tag for the category is relevant to the category by a confidence score for a corresponding one of the associated content tags of the set of images, a candidate event category score quantifying a likelihood that the set of images depict the category of event types;
• in response to the candidate event category score being over a predetermined threshold, determining a first candidate event type score quantifying a likelihood that the set of images belongs to a first event type in the category of event types;
• and
• in response to the first candidate event type score being over a predetermined first threshold, causing a user interface to present a representation that the set of images belongs to the first event type.

A media organization system comprising one or more hardware processors configured to cause the media organization system to perform operations comprising:

• identifying a collection of images having associated content tags and corresponding confidence scores;
• determining one or more candidate event types for the collection of images based on matching between the associated content tags of the images in the collection and a set of learned content tags representative of each of the one or more candidate event types;
• determining, for each candidate event type of the one or more candidate event types, based on weighting (i) a learned correlation score, that quantifies a likelihood that a learned content tag from the set of learned content tags for the candidate event type is relevant to the candidate event type, by (ii) a corresponding one of the confidence scores for the associated content tags, a candidate event type score indicative of a likelihood of the set of images belonging to the candidate event type;
• selecting an event type as one of the one or more candidate event types with a highest candidate event type score;
• and
• causing a user interface to organize the collection of images based on the selected event type.