UMAN_data

• The data of UMAN
• Learning Sequential Correlation for User Generated Textual Content Popularity Prediction
• link: https://drive.google.com/open?id=1QhOY80hXYE64LDyBm5vZoum7o4aaEAbH

Detail

• JSON format.
• keys:
  • nb_train, nb_vali, nb_test: number of training/validation/test data
  • train, vali, test: training/validation/test data
    • gpid: post id，increase with time.
    • label: label for prediction, the popularity scroe.
    • user: user id.    - time: Timestamps.
    • text: word id sequence.
  • nb_words, nb_items, nb_users: number of words/data/users
  • word_dict: mapping of word to word id
  • raw_text: raw text