Datasets:

Modalities:
Tabular
Text
Formats:
csv
Libraries:
Datasets
pandas
License:
JingyuanHe1222 commited on
Commit
2155f76
·
1 Parent(s): e29c412

update file structures

Browse files
README.md CHANGED
@@ -2,6 +2,31 @@
2
  license: mit
3
  ---
4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5
 
6
  # ClueWeb-Reco:
7
 
 
2
  license: mit
3
  ---
4
 
5
+ ---
6
+ configs:
7
+ - config_name: ordered_id_splits
8
+ data_files:
9
+ - split: train
10
+ path:
11
+ - "ordered_id_splits/valid_input.tsv"
12
+ - "ordered_id_splits/valid_target.tsv"
13
+ - "ordered_id_splits/test_input.tsv"
14
+ - split: test
15
+ path: "ordered_id_splits/ghi.csv"
16
+ default: true
17
+ - config_name: interaction_splits
18
+ data_files:
19
+ - split: valid
20
+ path:
21
+ - "interaction_splits/valid_inter_input.tsv"
22
+ - "interaction_splits/valid_inter_target.csv"
23
+ - split: test
24
+ path: "interaction_splits/test_input.tsv"
25
+ - config_name: clueweb_id_mapping
26
+ data_files: "cwid_to_id.tsv"
27
+ ---
28
+
29
+
30
 
31
  # ClueWeb-Reco:
32
 
cw_data_processing/ClueWeb22Api.py ADDED
@@ -0,0 +1,143 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import gzip
3
+ import zipfile
4
+
5
+ class ClueWeb22Api:
6
+
7
+ """
8
+ ClueWeb22Api adapted from source: https://github.com/lemurproject/ClueWeb22/blob/main/ClueWeb22Api.py
9
+ """
10
+
11
+ def __init__(self, cw22id, cw22root_path):
12
+ self.cw22id = cw22id
13
+ self.cw22root_path = cw22root_path
14
+
15
+ def get_base_filename_by_id(self, cw22id, cw22root_path, file_type='html'):
16
+ # /data/datasets/clueweb22/ClueWeb22_B/txt
17
+ html_path = self.cw22root_path + os.sep + file_type
18
+
19
+ # clueweb22-en00{subfolder_id}-{jjsongz_id}-{ddoc_id}"
20
+ id_parts = cw22id.split('-')
21
+ doc = int(id_parts[len(id_parts) - 1])
22
+
23
+ language = id_parts[1][:2] # en
24
+ segment = id_parts[1][:4] # en00
25
+ directory = id_parts[1] # en0000
26
+ # /data/datasets/clueweb22/ClueWeb22_B/txt + en + en00 + en0000
27
+ base_path = html_path + os.sep + language + os.sep + segment + os.sep + directory + os.sep
28
+ base_filename = base_path + id_parts[1] + '-' + id_parts[2]
29
+ return base_filename
30
+
31
+ def get_primary_node_ids(self, annotate_html):
32
+ annotations = annotate_html.annotations
33
+ primary_node_ids = []
34
+ for annotation in annotations:
35
+ if annotation.type == AnnotateHtml.AnnotationType.Primary:
36
+ primary_node_ids.append(int(annotation.nodeId))
37
+ primary_node_ids.sort()
38
+ return primary_node_ids
39
+
40
+ def get_html_from_warc(self):
41
+ cw22id = self.cw22id
42
+ cw22root_path = self.cw22root_path
43
+ base_filename = self.get_base_filename_by_id(cw22id, cw22root_path)
44
+
45
+ warc_path = base_filename + '.warc.gz'
46
+ offset_path = base_filename + '.warc.offset'
47
+
48
+ id_parts = cw22id.split('-')
49
+ doc = int(id_parts[len(id_parts) - 1])
50
+
51
+ #Get html from warc using offset
52
+ offset_length = len('{:010d}\n'.format(0, 0))
53
+ with open (warc_path,'rb') as f_warc:
54
+ with open (offset_path, 'r') as f_offset:
55
+ f_offset.seek(int(doc) * int(offset_length))
56
+ start_bytes = int (f_offset.read (offset_length).strip())
57
+ end_bytes = int (f_offset.read (offset_length).strip())
58
+ f_warc.seek(start_bytes)
59
+ record = f_warc.read(end_bytes - start_bytes)
60
+ record = gzip.decompress(record).decode('utf-8')
61
+
62
+ #Remove the WARC header to get the htmlStr
63
+ warc_header = ''
64
+ for line in record.splitlines():
65
+ warc_header += line
66
+ warc_header += '\r\n'
67
+ if len(line.strip()) == 0:
68
+ break
69
+ record = record[len(warc_header):]
70
+
71
+ return record
72
+
73
+
74
+
75
+ def get_json_record(self, record_type):
76
+ cw22id = self.cw22id
77
+ cw22root_path = self.cw22root_path
78
+ base_filename = self.get_base_filename_by_id(cw22id, cw22root_path, file_type=record_type)
79
+ json_path = base_filename + '.json.gz'
80
+ offset_path = base_filename + '.offset'
81
+
82
+ id_parts = cw22id.split('-')
83
+ doc = int(id_parts[len(id_parts) - 1])
84
+
85
+ offset_length = len('{:010d}\n'.format(0, 0))
86
+ with open (json_path,'rb') as f_json:
87
+ with open (offset_path, 'r') as f_offset:
88
+ f_offset.seek(int(doc) * int(offset_length))
89
+ start_bytes = int (f_offset.read (offset_length).strip())
90
+ end_bytes = int (f_offset.read (offset_length).strip())
91
+ f_json.seek(start_bytes)
92
+ record = f_json.read(end_bytes - start_bytes)
93
+ record = gzip.decompress(record).decode('utf-8')
94
+ return record
95
+
96
+
97
+ def get_clean_text(self):
98
+ record = self.get_json_record('txt')
99
+ return record
100
+
101
+ def get_inlinks(self):
102
+ record = self.get_json_record('inlink')
103
+ return record
104
+
105
+ def get_outlinks(self):
106
+ record = self.get_json_record('outlink')
107
+ return record
108
+
109
+
110
+ def get_marcoweb_clean_text(self):
111
+ record = self.get_marcoweb_json_record('txt')
112
+ return record
113
+
114
+ def get_marcoweb_json_record(self, record_type):
115
+ cw22id = self.cw22id
116
+ cw22root_path = self.cw22root_path
117
+
118
+ def get_base_filename():
119
+ html_path = cw22root_path + os.sep
120
+ id_parts = cw22id.split('-') # "clueweb22-en-{jjsongz_id}-{ddoc_id}"
121
+ language = id_parts[1]
122
+ base_path = html_path + os.sep + language + os.sep
123
+ base_filename = base_path + id_parts[1] + '-' + id_parts[2]
124
+ return base_filename
125
+
126
+ # custom file path
127
+ base_filename = get_base_filename()
128
+ json_path = base_filename + '.json.gz'
129
+ offset_path = base_filename + '.offset'
130
+
131
+ id_parts = cw22id.split('-')
132
+ doc = int(id_parts[len(id_parts) - 1])
133
+
134
+ offset_length = len('{:010d}\n'.format(0, 0))
135
+ with open (json_path,'rb') as f_json:
136
+ with open (offset_path, 'r') as f_offset:
137
+ f_offset.seek(int(doc) * int(offset_length))
138
+ start_bytes = int (f_offset.read (offset_length).strip())
139
+ end_bytes = int (f_offset.read (offset_length).strip())
140
+ f_json.seek(start_bytes)
141
+ record = f_json.read(end_bytes - start_bytes)
142
+ record = gzip.decompress(record).decode('utf-8')
143
+ return record
cw_data_processing/example_dataset.py ADDED
@@ -0,0 +1,56 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from ClueWeb22Api import ClueWeb22Api
2
+ from tqdm import tqdm
3
+
4
+
5
+ def get_clueweb_title(cwid, clueweb_path):
6
+ clueweb_api = ClueWeb22Api(cwid, clueweb_path)
7
+ clean_txt = eval(clueweb_api.get_clean_text())
8
+ content = clean_txt["Clean-Text"]
9
+ title = content.split('\n')[0].replace("\n", "").replace("\t", "").replace("\r", "").replace("\'", "").replace("\"", "").strip()
10
+ return title
11
+
12
+
13
+ def load_data_clueweb(filename, id_map_path, clueweb_path):
14
+ """
15
+ args:
16
+ filename: the validation set input (valid.tsv) or the test set input (test.tsv)
17
+ id_map_path: the mapping from official clueweb id to our internal clueweb id (cwid_to_id.tsv)
18
+ clueweb_path: your clueweb dataset path on disk (ex. /../data/datasets/clueweb22/ClueWeb22_B)
19
+ """
20
+
21
+ # load mapping to official clueweb doc ids
22
+ id_to_cwid = {}
23
+ with open(id_map_path, "r") as f:
24
+ for line in f:
25
+ parts = line.strip().split("\t")
26
+ id_to_cwid[int(parts[1])] = parts[0]
27
+
28
+
29
+ data = {}
30
+ lines = open(filename, 'r').readlines()
31
+ for line in tqdm(lines[1:]):
32
+ history_titles = list()
33
+ line = line.strip().split('\t')
34
+
35
+ session_id = line[0]
36
+ history = line[1].split(",")
37
+
38
+ for internal_id in history:
39
+ internal_id = int(internal_id)
40
+ # get the title of the item by ClueWebApi
41
+ title = get_clueweb_title(id_to_cwid[internal_id], clueweb_path)
42
+ history_titles.append(title)
43
+
44
+ data[session_id] = history_titles
45
+
46
+ return data
47
+
48
+
49
+ id_map_path="ClueWeb-Reco/cwid_to_id.tsv"
50
+ seq_data_path="ClueWeb-Reco_public/ordered_id_splits/valid_input.tsv"
51
+ clueweb_path="/data/datasets/clueweb22/ClueWeb22_B"
52
+
53
+ # dictionary with session_ids as key, list of strings of the titles of historically interacted item
54
+ data = load_data_clueweb(filename=seq_data_path, id_map_path=id_map_path, clueweb_path=clueweb_path)
55
+
56
+
test_inter_input.tsv → interaction_splits/test_inter_input.tsv RENAMED
File without changes
valid_inter_input.tsv → interaction_splits/valid_inter_input.tsv RENAMED
File without changes
valid_inter_target.tsv → interaction_splits/valid_inter_target.tsv RENAMED
File without changes
test_input.tsv → ordered_id_splits/test_input.tsv RENAMED
File without changes
valid_input.tsv → ordered_id_splits/valid_input.tsv RENAMED
File without changes
valid_target.tsv → ordered_id_splits/valid_target.tsv RENAMED
File without changes