Automatic generation of highlight clips for events

A computer-implemented method, comprising:

• receiving, by a computer system, unstructured data associated with an event;
• obtaining, by the computer system, audio data and video data for the event;
• extracting, by the computer system, structured data from the unstructured data using one or more feature extractors, the one or more feature extractors configured to be utilized with a type of data associated with a portion of the unstructured data;
• determining, by the computer system, a duration for a highlight clip of the event based at least in part on a model that uses the structured data, the audio data and the video data of the event;
• generating, by the computer system, metadata that identifies the duration for the highlight clip of the event based at least in part on the model, the model configured to utilize one or more thresholds for comparing each different type of the structured data to generate the metadata, and wherein a threshold of the one or more thresholds is selected by the model based at least in part on a type of the event;
• storing, by the computer system, the highlight clip for the event based at least in part on the metadata;
• and
• transmitting, by the computer system and to a user device, the highlight clip for presentation via a user interface of the user device and within a certain time period of receiving the unstructured data.

A computer system comprising:

• a memory configured to store computer-executable instructions;
• and
• a processor in communication with the memory configured to execute the computer-executable instructions to at least:
• receive unstructured data associated with an event;
• obtain audio data and video data for the event;
• extract structured data from the unstructured data using one or more feature extractors, the one or more feature extractors configured to be utilized with a type of data associated with a portion of the unstructured data;
• and
• generate metadata that identifies a duration for a highlight clip of the event based at least in part on a model that uses the structured data, the audio data and the video data of the event, and one or more thresholds for comparing each different type of the structured data, wherein a threshold of the one or more thresholds is selected by the model based at least in part on a type of the event.