A system and method for determining a similarity measure between a first network broadcast time segment and a second network broadcast time segment. The method includes generating first training data comprising groupings of similar broadcast networks and groupings of dissimilar broadcast networks, extracting second training data comprising audience data and genre data for time segments of network broadcasts, and training an encoder neural network based on the first and second training data. The method includes executing the trained neural network with input data comprising the first network broadcast time segment and determining a similarity measure relative to the second network broadcast time segment.