How to train a model to categorize user browser history?

Part-1: How do we constitute serialized training data for model training? For instance, if I were creating a prediction model using visited web pages data, what’s the best way to render such data?

Example from training data: [<website 1, website 2, website 3>, entertainment], where entertainment is one of the labels (Summarizing the category of the web pages that user browsed in the last session)

Part-2: Is RNN the best method to train on such data for classification?