tools/accuracy_checker/accuracy_checker/annotation_converters/ncf_converter.py

   1 """
   2 Copyright (c) 2019 Intel Corporation
   3
   4 Licensed under the Apache License, Version 2.0 (the "License");
   5 you may not use this file except in compliance with the License.
   6 You may obtain a copy of the License at
   7
   8       http://www.apache.org/licenses/LICENSE-2.0
   9
  10 Unless required by applicable law or agreed to in writing, software
  11 distributed under the License is distributed on an "AS IS" BASIS,
  12 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13 See the License for the specific language governing permissions and
  14 limitations under the License.
  15 """
  16
  17
  18 from ..representation import HitRatioAnnotation
  19 from ..utils import read_txt, get_path
  20 from ..config import PathField, NumberField
  21
  22 from .format_converter import BaseFormatConverter, BaseFormatConverterConfig
  23
  24
  25 class NCFDatasetConverterConfig(BaseFormatConverterConfig):
  26     raiting_file = PathField()
  27     negative_file = PathField()
  28     users_max_number = NumberField(optional=True)
  29
  30
  31 class NCFConverter(BaseFormatConverter):
  32     __provider__ = "ncf_converter"
  33
  34     _config_validator_type = NCFDatasetConverterConfig
  35
  36     def configure(self):
  37         self.raiting_file = self.config['raiting_file']
  38         self.negative_file = self.config['negative_file']
  39         if 'users_max_number' in self.config:
  40             self.users_max_number = self.config['users_max_number']
  41         else:
  42             self.users_max_number = -1
  43
  44     def convert(self):
  45         annotations = []
  46         users = []
  47
  48         for file_row in read_txt(self.raiting_file):
  49             user_id, item_id, _ = file_row.split()
  50             users.append(user_id)
  51             identifier = ['u:'+user_id, 'i:' + item_id]
  52             annotations.append(HitRatioAnnotation(identifier))
  53             if self.users_max_number > 0 and len(users) >= self.users_max_number:
  54                 break;
  55
  56         item_numbers = 1
  57
  58         items_neg = []
  59         with get_path(self.negative_file).open() as content:
  60             for file_row in content:
  61                 items = file_row.split()
  62                 items_neg.append(items)
  63                 if self.users_max_number > 0 and len(items_neg) >= self.users_max_number:
  64                     break;
  65
  66         if items_neg:
  67             iterations = len(items_neg[0])
  68             item_numbers += iterations
  69             for i in range(iterations):
  70                 for user in users:
  71                     item = items_neg[int(user)][i]
  72                     identifier = ['u:' + user, 'i:' + item]
  73                     annotations.append(HitRatioAnnotation(identifier, False))
  74
  75         return annotations, {'users_number': len(users), 'item_numbers': item_numbers}