guenthermi commited on
Commit
705db86
1 Parent(s): 4b180be

support multiple languages

Browse files
Files changed (2) hide show
  1. README.md +5 -0
  2. xmarket_ml.py +150 -0
README.md ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ # XMarket Category to Product Retrieval Dataset
2
+
3
+ An ecommerce category to product retrieval dataset in multiple languages.
4
+ The data comes from:
5
+ https://xmrec.github.io/
xmarket_ml.py ADDED
@@ -0,0 +1,150 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """XMarket retrieval dataset for multiple languages"""
2
+
3
+
4
+ import json
5
+ import gzip
6
+
7
+ from typing import List
8
+
9
+ import datasets
10
+
11
+ _DESCRIPTION = """\
12
+ An ecommerce category to product retrieval dataset in multiple languages.
13
+ The data comes from:
14
+ https://xmrec.github.io/
15
+ """
16
+
17
+ _HOMEPAGE_URL = 'https://xmrec.github.io/'
18
+ _LANGUAGES = {'es': 'ES', 'de': 'DE', 'en': 'EN'}
19
+ _VERSION = '1.0.0'
20
+
21
+
22
+ URLS = [
23
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Arts_Crafts_and_Sewing/metadata_%s_Arts_Crafts_and_Sewing.json.gz',
24
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Automotive/metadata_%s_Automotive.json.gz',
25
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Books/metadata_%s_Books.json.gz',
26
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/CDs_and_Vinyl/metadata_%s_CDs_and_Vinyl.json.gz',
27
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Cell_Phones_and_Accessories/metadata_%s_Cell_Phones_and_Accessories.json.gz',
28
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Digital_Music/metadata_%s_Digital_Music.json.gz',
29
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Electronics/metadata_%s_Electronics.json.gz',
30
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Grocery_and_Gourmet_Food/metadata_%s_Grocery_and_Gourmet_Food.json.gz',
31
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Home_and_Kitchen/metadata_%s_Home_and_Kitchen.json.gz',
32
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Industrial_and_Scientific/metadata_%s_Industrial_and_Scientific.json.gz',
33
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Kindle_Store/metadata_%s_Kindle_Store.json.gz',
34
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Movies_and_TV/metadata_%s_Movies_and_TV.json.gz',
35
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Musical_Instruments/metadata_%s_Musical_Instruments.json.gz',
36
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Office_Products/metadata_%s_Office_Products.json.gz',
37
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Sports_and_Outdoors/metadata_%s_Sports_and_Outdoors.json.gz',
38
+ 'https://ciir.cs.umass.edu/downloads/XMarket/FULL/%s/Toys_and_Games/metadata_%s_Toys_and_Games.json.gz',
39
+ ]
40
+
41
+
42
+ class XMarketMLConfig(datasets.BuilderConfig):
43
+ """BuilderConfig for XMarketES."""
44
+
45
+ def __init__(self, languages=None, **kwargs):
46
+ super(XMarketMLConfig, self).__init__(
47
+ version=datasets.Version(_VERSION, ''), **kwargs
48
+ )
49
+ self.languages = languages
50
+
51
+
52
+ class XMarketML(datasets.GeneratorBasedBuilder):
53
+ """The XMarketES category to product retrieval dataset"""
54
+
55
+ BUILDER_CONFIGS = [
56
+ XMarketMLConfig(
57
+ name=name,
58
+ description=f'{name.title()} of the multilingual XMarket dataset.',
59
+ languages=None,
60
+ )
61
+ for name in ['corpus', 'queries', 'qrels']
62
+ ]
63
+
64
+ BUILDER_CONFIG_CLASS = XMarketMLConfig
65
+
66
+ def __init__(self, *args, **kwargs):
67
+ super().__init__(*args, **kwargs)
68
+ self._data = None
69
+
70
+ def _info(self):
71
+ return datasets.DatasetInfo(
72
+ description=_DESCRIPTION,
73
+ features=datasets.Features(
74
+ {
75
+ "_id": datasets.Value("string"),
76
+ "title": datasets.Value("string"),
77
+ "text": datasets.Value("string"),
78
+ }
79
+ ),
80
+ supervised_keys=None,
81
+ homepage=_HOMEPAGE_URL,
82
+ )
83
+
84
+ def _split_generators(self, dl_manager: datasets.DownloadManager):
85
+ json_files = []
86
+ for lang in self.config.languages:
87
+ for url in URLS:
88
+ json_files.append(dl_manager.download(url % (lang, lang)))
89
+ return [
90
+ datasets.SplitGenerator(
91
+ name=datasets.Split.TEST,
92
+ gen_kwargs={
93
+ 'files': json_files,
94
+ 'split': 'test',
95
+ },
96
+ ),
97
+ ]
98
+
99
+ def _generate_examples(
100
+ self,
101
+ files: List[str],
102
+ split: str = None,
103
+ ):
104
+ if not self._data:
105
+ corpus = []
106
+ queries = dict()
107
+ qrels = dict()
108
+ for file in files:
109
+ with gzip.open(file, 'rt') as f:
110
+ for line in f:
111
+ product = json.loads(line)
112
+ corpus.append(
113
+ {
114
+ "_id": product['asin'],
115
+ "title": product['title'],
116
+ "text": product['description'],
117
+ }
118
+ )
119
+ for category in product['categories']:
120
+ if category in queries:
121
+ qid = queries[category]
122
+ qrels[qid].append(product['asin'])
123
+ else:
124
+ qid = f'q{len(queries)}'
125
+ queries[category] = qid
126
+ qrels[qid] = [product['asin']]
127
+ self._data = {
128
+ 'corpus': corpus,
129
+ 'queries': queries,
130
+ 'qrels': qrels,
131
+ }
132
+ if self.config.name == 'corpus':
133
+ for line in self._data['corpus']:
134
+ yield line['_id'], line
135
+ elif self.config.name == 'queries':
136
+ for query, qid in queries.items():
137
+ yield qid, {
138
+ "_id": qid,
139
+ "title": '',
140
+ "text": query,
141
+ }
142
+ elif self.config.name == 'qrels':
143
+ for qid, dids in qrels.items():
144
+ yield qid, {
145
+ "_id": qid,
146
+ "title": '',
147
+ "text": ' '.join(dids),
148
+ }
149
+ else:
150
+ raise ValueError(f'Unknown config name: {self.config.name}')