calI2I.py 3.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122
  1. #coding utf-8
  2. import sys
  3. from operator import itemgetter
  4. import json
  5. if __name__=="__main__":
  6. #1.load data
  7. nowdate=sys.argv[1]
  8. f = open("./data/user_item_share_"+nowdate)
  9. user_item_dict={}
  10. item_dict = {}
  11. while True:
  12. line = f.readline()
  13. if not line:
  14. break
  15. items = line.strip().split("\t")
  16. if len(items)<3:
  17. continue
  18. key = (items[1],items[2])
  19. #print(key)
  20. if key not in user_item_dict:
  21. user_item_dict[key] = 1
  22. else:
  23. user_item_dict[key] = user_item_dict[key]+1
  24. if items[2] not in item_dict:
  25. item_dict[items[2]] = 1
  26. else:
  27. item_dict[items[2]] = item_dict[items[2]]+1
  28. f.close()
  29. #((user,item), score)
  30. #print(user_item_dict)
  31. #2. (uid, [(vid, score)....])
  32. user_group_dict = {}
  33. for k, v in user_item_dict.items():
  34. uid = k[0]
  35. vid = k[1]
  36. score = v
  37. vid_list = []
  38. if uid not in user_group_dict:
  39. vid_list.append((vid, score))
  40. user_group_dict[uid] = vid_list
  41. else:
  42. vid_list = user_group_dict[uid]
  43. vid_list.append((vid, score))
  44. user_group_dict[uid] = vid_list
  45. #print(user_group_dict)
  46. item_pair_dict = {}
  47. #3. expand item
  48. for k, v_list in user_group_dict.items():
  49. v_n = len(v_list)
  50. if v_n<2:
  51. continue
  52. for i in range(v_n):
  53. for j in range(1, v_n):
  54. if v_list[i][0] == v_list[j][0]:
  55. continue
  56. item_key = (v_list[i][0], v_list[j][0])
  57. item_score = min(v_list[i][1], v_list[j][1])
  58. if item_key not in item_pair_dict:
  59. item_pair_dict[item_key] = item_score
  60. else:
  61. item_pair_dict[item_key] = item_pair_dict[item_key]+item_score
  62. #print(item_pair_dict)
  63. print(len(item_pair_dict))
  64. print(len(item_dict))
  65. left_pair_num = 0
  66. rec_item_dict = {}
  67. #4. rec item
  68. for k, v in item_pair_dict.items():
  69. if v<2:
  70. continue
  71. left_pair_num+=1
  72. item1 = k[0]
  73. item2 = k[1]
  74. pair_score = v
  75. if item1 in item_dict:
  76. item_score1 = item_dict[item1]
  77. i2i_pro = pair_score/(item_score1+5)
  78. rec_list1 = []
  79. if item2 not in rec_item_dict:
  80. rec_list1.append((item1, i2i_pro, pair_score, item_score1))
  81. rec_item_dict[item2] = rec_list1
  82. else:
  83. rec_list1 = rec_item_dict[item2]
  84. rec_list1.append((item1, i2i_pro, pair_score, item_score1))
  85. rec_item_dict[item2] = rec_list1
  86. if item2 in item_dict:
  87. item_score2 = item_dict[item2]
  88. i2i_pro = pair_score/(item_score2+5)
  89. rec_list2 = []
  90. if item1 not in rec_item_dict:
  91. rec_list2.append((item2, i2i_pro, pair_score, item_score2))
  92. rec_item_dict[item1] = rec_list2
  93. else:
  94. rec_list2 = rec_item_dict[item1]
  95. rec_list2.append((item2, i2i_pro, pair_score, item_score2))
  96. rec_item_dict[item1] = rec_list2
  97. #(item, share_count)
  98. print(left_pair_num)
  99. #print(rec_item_dict)
  100. final_rec_list = []
  101. #f = open("rec_result", "w")
  102. #5. sorted item_list
  103. for k,v in rec_item_dict.items():
  104. v_set = set('')
  105. value_list = v
  106. dup_list = []
  107. for item in value_list:
  108. if item[0] in v_set:
  109. continue
  110. v_set.add(item[0])
  111. dup_list.append(item)
  112. sorted_v = sorted(dup_list, key=itemgetter(1), reverse=True)
  113. final_rec_list.append((k, sorted_v))
  114. #print(final_rec_list[:1])
  115. #json_str = json.dumps(final_rec_list)
  116. with open("./data/rec_result_"+nowdate+".json", "w") as f :
  117. json.dump(final_rec_list, f)