-
Notifications
You must be signed in to change notification settings - Fork 8
/
constants.py
285 lines (265 loc) · 5.73 KB
/
constants.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
# -*- coding: utf-8 -*-
import re
from itertools import chain
from os.path import join
# --- default constants definitions ---
# default paths
# DATA_BASE = "../../master_cloud/corpora"
DATA_BASE = "../data"
etl_base = "preprocessed"
ETL_PATH = join(DATA_BASE, etl_base)
# TODO: add local path to arguments/options
LOCL_PATH = etl_base
FULL_PATH = join(DATA_BASE, LOCL_PATH)
nlp_base = "preprocessed/nlp"
NLP_PATH = join(DATA_BASE, nlp_base)
smpl_base = "preprocessed/simple"
SMPL_PATH = join(DATA_BASE, smpl_base)
tmp_base = "preprocessed/tmp"
TMP_PATH = join(DATA_BASE, tmp_base)
SPCY_PATH = join(NLP_PATH, "spacy_model")
VOC_PATH = join(SPCY_PATH, "vocab")
LDA_PATH = join(ETL_PATH, "LDAmodel")
LSI_PATH = join(ETL_PATH, "LSImodel")
EMB_PATH = join(ETL_PATH, "embeddings")
TPX_PATH = join(LDA_PATH, "noun", "bow", "topics")
# data scheme
DATASET = "dataset"
SUBSET = "subset"
TIME = "date_time"
ID = "doc_id"
ID2 = "doc_subid"
TITLE = "title"
# AUTHOR -> mostly unknown or pseudonym
# SUBTITLE -> use DESCRIPTION
# CATEGORY -> use DESCRIPTION or LINKS
META = [DATASET, SUBSET, TIME, ID, ID2, TITLE]
TEXT = "text"
# The DESCRIPTION and LINKS fields were introduced with dewiki and are so far unused in the other
# datasets.
# DESCRIPTION: Would be nice to add this especially to the news sites datasets.
# In dewiki it contains rather a subtitle than a description.
DESCR = "description"
# LINKS: could be used to link forum threads together, although this is probably already done via ID[2]
LINKS = "links"
# moved from META to DATA => hashes have to be recalculated!
TAGS = "tags"
DATA = [DESCR, TEXT, LINKS, TAGS]
HASH = "hash"
TOKEN = "token"
# --- additional constants
# Universal Tagset
ADJ = "ADJ"
ADV = "ADV"
INTJ = "INTJ"
NOUN = "NOUN"
PROPN = "PROPN"
VERB = "VERB"
ADP = "ADP"
AUX = "AUX"
CCONJ = "CCONJ"
CONJ = "CONJ"
DET = "DET"
NUM = "NUM"
PART = "PART"
PRON = "PRON"
SCONJ = "SCONJ"
PUNCT = "PUNCT"
SYM = "SYM"
X = "X"
# additional
SPACE = "SPACE"
PHRASE = "PHRASE"
NER = "NER"
NPHRASE = "NPHRASE"
# keys
IWNLP = "IWNLP"
POS = "POS"
TOK_IDX = "tok_idx"
START = "start"
LEMMA = "lemma"
TAG = "tag"
STOP = "stop"
ENT_TYPE = "ent_type"
ENT_IOB = "ent_iob"
ENT_IDX = "ent_idx"
KNOWN = "known"
SENT_IDX = "sent_idx"
SENT_START = "sent_start"
NOUN_PHRASE = "noun_phrase"
# --- for LDA modeling
# available datasets (unique)
DATASETS = {
"dewac": "dewac",
"dewac1": "dewac1",
"dewiki": "dewiki",
"E": "Europarl",
"FA": "FAZ",
"FO": "FOCUS",
"N": "news",
"O": "OnlineParticipation",
"P": "PoliticalSpeeches",
"S": "speeches",
}
# additional keys for available datasets
DSETS = DATASETS.copy()
DSETS.update(
{
"dewi": "dewiki",
"dewik": "dewiki",
"dewa": "dewac",
"dewa1": "dewac1",
"e": "Europarl",
"europarl": "Europarl",
"fa": "FAZ",
"faz": "FAZ",
"fo": "FOCUS",
"focus": "FOCUS",
"o": "OnlineParticipation",
"onlineparticipation": "OnlineParticipation",
"p": "PoliticalSpeeches",
"politicalspeeches": "PoliticalSpeeches",
"n": "news",
"f": "news",
"F": "news",
"s": "speeches",
}
)
METRICS = ("ref", "u_mass", "c_v", "c_uci", "c_npmi", "vote")
PARAMS = ("a42", "b42", "c42", "d42", "e42")
NBTOPICS = (10, 25, 50, 100)
VERSIONS = ("noun", "noun-verb", "noun-verb-adj")
CORPUS_TYPE = ("bow", "tfidf")
# --- filter lookup table
# the follwoing tokens are filtered befor applying LDA training
BAD_TOKENS_DICT = {
"Europarl": [
"E.",
"Kerr",
"The",
"la",
"ia",
"For",
"Ieke",
"the",
"WPA",
"INSPIRE",
"EN",
"ASEM",
"ISA",
"EIT",
],
"FAZ_combined": [
"S.",
"B.",
"P.",
"of",
],
"FOCUS_cleansed": [
"OTS",
"RSS",
"of",
"UP",
"v.",
],
"OnlineParticipation": [
"Re",
"@#1",
"@#2",
"@#3",
"@#4",
"@#5",
"@#6",
"@#7",
"@#8",
"@#9",
"@#1.1",
"Für",
"Muss",
"etc",
"sorry",
"Ggf",
"u.a.",
"z.B.",
"B.",
"stimmt",
";-)",
"lieber",
"o.",
"Ja",
"Desweiteren",
"@#4.1.1",
],
"PoliticalSpeeches": [
"ZIF",
"of",
"and",
"DFFF",
],
"dewiki": [],
"dewac": [
"H.",
"m.",
"W.",
"K.",
"g.",
"r.",
"A.",
"f.",
"l.",
"J.",
"EZLN",
"LAGH",
"LSVD",
"AdsD",
"NAD",
"DÖW",
"Rn",
],
}
BAD_TOKENS = set(chain(*BAD_TOKENS_DICT.values()))
PLACEHOLDER = "[[PLACEHOLDER]]"
MINIMAL_PATTERN = re.compile(r".\.")
# NOUN_PATTERN = re.compile(r'^([0-9]+.*?)*?[A-Za-zÄÖÜäöüß].*?[A-Za-zÄÖÜäöüß0-9].*')
NOUN_PATTERN = re.compile(r"^([0-9]+.*?)*?[A-Za-zÄÖÜäöüß].*")
POS_N = [NOUN, PROPN, NER, NPHRASE]
POS_NV = [NOUN, PROPN, NER, NPHRASE, VERB]
POS_NVA = [NOUN, PROPN, NER, NPHRASE, VERB, ADJ, ADV]
"""
list of tokens to ignore when at the beginning of a phrase
This is needed to avoid changing all appearances of for example
'die Firma' to 'Die_Firma' since this is also a movie title.
"""
BAD_FIRST_PHRASE_TOKEN = {
"ab",
"seit",
"in",
"der",
"die",
"das",
"an",
"am",
"diese",
"bis",
"ein",
"es",
"mit",
"im",
"für",
"zur",
"auf",
"!",
"(",
"ich",
"so",
"auch",
"wir",
"auch",
"mich",
"du",
}
GOOD_IDS = {
"dewac": join(ETL_PATH, "dewac_good_ids.pickle"),
"dewiki": join(ETL_PATH, "dewiki_good_ids.pickle"),
}