video_rank.py 2.8 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283
  1. import random
  2. from log import Log
  3. from config import set_config
  4. log_ = Log()
  5. config_ = set_config()
  6. def video_rank(data, size):
  7. """
  8. 视频分发排序
  9. :param data: 各路召回的视频 type-dict {'rov_pool_recall': [], 'flow_pool_recall': []}
  10. :param size: 请求数
  11. :return: rank_result
  12. """
  13. # 将各路召回的视频按照score从大到小排序
  14. # ROV召回池
  15. rov_recall_rank = sorted(data['rov_pool_recall'], key=lambda k: (k.get('rovScore'), 0), reverse=True)
  16. # 流量池
  17. flow_recall_rank = sorted(data['flow_pool_recall'], key=lambda k: (k.get('rovScore'), 0), reverse=True)
  18. # 对各路召回的视频进行去重
  19. rov_recall_rank, flow_recall_rank = remove_duplicate(rov_recall=rov_recall_rank, flow_recall=flow_recall_rank)
  20. # 从ROV召回池中获取top k
  21. if len(rov_recall_rank) > 0:
  22. rank_result = rov_recall_rank[:config_.K]
  23. rov_recall_rank = rov_recall_rank[config_.K:]
  24. elif len(flow_recall_rank) > 0:
  25. rank_result = flow_recall_rank[:config_.K]
  26. flow_recall_rank = flow_recall_rank[config_.K:]
  27. else:
  28. # 兜底策略
  29. return None
  30. if not rov_recall_rank and not flow_recall_rank:
  31. # 兜底策略
  32. return None
  33. # 按概率 p 及score排序获取 size - k 个视频
  34. i = 0
  35. while i < size - config_.K:
  36. # 随机生成[0, 1)浮点数
  37. rand = random.random()
  38. if rand < config_.P:
  39. if flow_recall_rank:
  40. rank_result.append(flow_recall_rank[0])
  41. flow_recall_rank.remove(flow_recall_rank[0])
  42. else:
  43. rank_result.append(rov_recall_rank[:size - config_.K - i])
  44. return rank_result
  45. else:
  46. if rov_recall_rank:
  47. rank_result.append(rov_recall_rank[0])
  48. rov_recall_rank.remove(rov_recall_rank[0])
  49. else:
  50. rank_result.append(flow_recall_rank[:size - config_.K - i])
  51. return rank_result
  52. i += 1
  53. return rank_result
  54. def remove_duplicate(rov_recall, flow_recall):
  55. """
  56. 对多路召回的视频去重
  57. 去重原则:
  58. 如果视频在ROV召回池topK,则保留ROV召回池,否则保留流量池
  59. :param rov_recall: ROV召回池-已排序
  60. :param flow_recall: 流量池-已排序
  61. :return:
  62. """
  63. flow_recall_remove = []
  64. flow_recall_video_ids = [item[0] for item in flow_recall]
  65. # rov_recall topK
  66. for item in rov_recall[:config_.K]:
  67. if item[0] in flow_recall_video_ids:
  68. flow_recall_remove.append(item[0])
  69. # other
  70. for item in rov_recall[config_.K:]:
  71. if item[0] in flow_recall_video_ids:
  72. rov_recall.remove(item)
  73. # flow recall remove
  74. for item in flow_recall:
  75. if item[0] in flow_recall_remove:
  76. flow_recall.remove(item)
  77. return rov_recall, flow_recall