-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathNLP_preprocessing pipeline.R
563 lines (467 loc) · 24.5 KB
/
NLP_preprocessing pipeline.R
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
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
#NLP Pipeline
# initial creation date: 03/11/2021
# version: 1.54
# current version date: 11/07/2024
# author: Carlo Santagiustina
# email: [email protected]
# Please do not share this code with people outside the DMI 2024 Summer School group
options(encoding='UTF-8')
# Load libraries
#Rstudio should ask you to install it
install.packages(c("devtools","tidyverse","quanteda","newsmap","ggplot2","stm","spacyr","visNetwork"))
devtools::install_github("hadley/emo")
devtools::install_github("richfitz/remoji")
require(quanteda)
require(newsmap)
require(ggplot2)
require(spacyr)
require(remoji)
require(emo)
require(tidyverse)
library(readr)
library(visNetwork)
#Import data using the readr library
# data=readr::read_csv(file = "[filename].csv",col_types = cols(id = col_character()
#)
#
# )
data$language_descr=textcat::textcat(data$text)# infer language using textcat
table(data$language_descr) %>% sort(decreasing = T)
dict_emoji=remoji:::dat_core$emoji
names(dict_emoji)=remoji:::dat_core$description
data=as.data.frame(data)
#data=data[!is.na(data$text),] # remove rows with empty text
rownames(data) =data$id
data=data[, !(colnames(data) %in% colnames(data)[duplicated(colnames(data))])]
data=data[is.na(data$retweeted_id),]# keep only original tweets
data_clean= data %>% #remove emojis
dplyr::mutate(emoji = emo::ji_match_all(user_description)) %>% #remove all strange characters (emoji, etc)
dplyr::mutate(clean_text = iconv(
user_description,
to = "latin1",
from = "UTF-8" ,
sub = ' '
)) %>% #clean & character
dplyr::mutate(clean_text = gsub(
"& ", #this is the & character in HTML
"and ",
clean_text,
ignore.case = FALSE,
perl = TRUE
)) %>% #clean & character
dplyr::mutate(clean_text = gsub(
" &",
" and",
clean_text,
ignore.case = FALSE,
perl = TRUE
))%>% #clean & character
dplyr::mutate(clean_text = gsub(
"&",
"&",
clean_text,
ignore.case = FALSE,
perl = TRUE
)) %>% #clean & character
dplyr::mutate(clean_text = gsub(
"Q&A",
"Q&A",
clean_text,
ignore.case = FALSE,
perl = TRUE
)) %>% #extract urls
dplyr::mutate(text_urls = regmatches(
clean_text,
gregexpr(
"(http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\\(\\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+)",
#"(http[s]?:?\\/?\\/?t?\\.?c?o?\\/?[A-Za-z0-9\\.\\/]{0,}[^ [:punct:]])",
clean_text,
perl = TRUE
)
)) %>% #remove urls
dplyr::mutate(
clean_text = gsub(
"http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\\(\\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+",
#"(http[s]?:?\\/?\\/?t?\\.?c?o?\\/?[A-Za-z0-9\\.\\/]{0,}[^ [:punct:]])",
" ",
clean_text,
perl = TRUE
)
) %>% #extract #tags
dplyr::mutate(text_hashtags = regmatches(
clean_text,
gregexpr(
"([#]{1}[a-zA-Z0-9_]{1,})",
clean_text,
perl = TRUE
)
)) %>%
#remove # characters of from words
dplyr::mutate(clean_text = sapply(
regmatches(
clean_text,
gregexpr("([#]{1})(?=[a-zA-Z0-9_]{1,})", clean_text, perl = TRUE),
invert = TRUE
),
paste0,
collapse = " "
)) %>% #extract $tags
dplyr::mutate(text_fintags = regmatches(
clean_text,
gregexpr(
"([$]{1}[a-zA-Z]{2,8})",
clean_text,
perl = TRUE
)
)) %>% #remove $ characters of finance tags
dplyr::mutate(clean_text = sapply(
regmatches(
clean_text,
gregexpr("([$]{1})(?=[A-Za-z]{2,8})", clean_text, perl = TRUE),
invert = TRUE
),
paste0,
collapse = " "
))%>% #remove @ characters of user tags
dplyr::mutate(clean_text = sapply(
regmatches(
clean_text,
gregexpr("([@]{1})(?=[ ]?[A-Za-z0-9]{2,8})", clean_text, perl = TRUE),
invert = TRUE
),
paste0,
collapse = " "
))%>% #extract standard character emoticones
dplyr::mutate(text_emoticones = regmatches(
clean_text,
gregexpr(pattern = "([>]{0,1}[:;=][']{0,1}[c\\-? ?\\^]{0,1}[()\\]\\|][()\\]\\|]?)|(\\s[3>]?[;=X:8][']{0,1}[c\\-? ?\\^]?[()\\]SsbpPDBOo0\\|][()\\]\\|]?\\s)", clean_text, perl = TRUE)
)) %>% #remove emoticones
dplyr::mutate(
clean_text = gsub(pattern = "([>]{0,1}[:;=][']{0,1}[c\\-? ?\\^]{0,1}[()\\]\\|][()\\]\\|]?)|(\\s[3>]?[;=X:8][']{0,1}[c\\-? ?\\^]?[()\\]SsbpPDBOo0\\|][()\\]\\|]?\\s)", " ", clean_text, perl = TRUE)
)
saveRDS(data_clean,file = "processed_data.RData")
data=readRDS("processed_data_tweets.RData")
#### NLP pipeline for NER and for extracting dep-rel ####
library(spacyr)# pacchetto utilizzato
spacyr::spacy_install()# run it only once
spacyr::spacy_download_langmodel(lang_models = "fr_core_news_md")# download french model run it only once
mySpacyrConsolidateEnt=function (x, concatenator = "_")
{
library(tidyverse)
spacy_result <- data.table::as.data.table(x)
lemma_space <- token_space <- NULL
entity <- entity_type <- entity_count <- iob <- entity_id <- .N <- .SD <- `:=` <- token <- lemma <- pos <- tag <- new_token_id <- token_id <- sentence_id <- doc_id <- NULL
if (!"entity" %in% names(spacy_result)) {
stop("no entity in parsed object: rerun spacy_parse() with entity = TRUE")
}
spacy_result[, `:=`(entity_type, sub("_.+", "", entity))]
spacy_result[, `:=`(iob, sub(".+_", "", entity))]
extended_list <- c("DATE", "TIME", "PERCENT", "MONEY", "QUANTITY",
"ORDINAL", "CARDINAL")
spacy_result[, `:=`(entity_count, ifelse(iob == "B" | iob ==
"", 1, 0))]
spacy_result[, `:=`(entity_id, cumsum(entity_count)),
by = c("doc_id", "sentence_id")]
spacy_result[, `:=`(token_space, token)]
spacy_result[, `:=`(lemma_space, lemma)]
spacy_result_modified = spacy_result %>%
group_by(doc_id, sentence_id, entity_id) %>%
mutate(token_entity= paste0(token_space, collapse = concatenator))
spacy_result_modified = spacy_result_modified %>%
group_by(doc_id, sentence_id, entity_id) %>%
mutate(lemma_entity= paste0(lemma_space, collapse = concatenator))
spacy_result_modified$ent_root=grepl(pattern = "B",x = spacy_result_modified$iob,perl = T)
spacy_result_modified= spacy_result_modified[spacy_result_modified$ent_root==T | (spacy_result_modified$iob %in% c("")),]
ret <- as.data.frame(spacy_result_modified)
class(ret) <- c("spacyr_parsed", class(ret))
ret
}
#spacyr::spacy_install(envname = "spacy_condaenv",version = "latest")#installa conda su pc
spacy_initialize('fr_core_news_md',condaenv = "~/Documents/Miniconda/envs/spacyr/")
data_bkp=data_clean
data_clean$doc_id=paste0("text",1:nrow(data_clean))#check this!
#data$text=gsub("ndr"," ",x = data$text,ignore.case = T)
rownames(data_clean)=data_clean$doc_id
saveRDS(data_clean,file = "data_before_NLP_text.RDS")
NLP_data_raw=spacy_parse(data_clean$clean_text, dependency = TRUE, pos = TRUE,entity = T,multithread = T) #,"morph"
spacy_finalize()
saveRDS(NLP_data_raw,"raw_data_tweets.RDS")
NLP_data_entities=mySpacyrConsolidateEnt(NLP_data_raw,concatenator = " ")
saveRDS(NLP_data_entities,"NLP_data_entities_tweets.RDS")
NLP_data_entities=NLP_data_entities[!((NLP_data_entities$pos %in% c("PUNCT","SPACE","SYM","X"))),]#"AUX","CCONJ","SYM","PRON","DET","INTJ","PART","ADP","SCONJ","X"
NLP_data_entities=NLP_data_entities[grepl("[[:alnum:]]{1,}",x = NLP_data_entities$token_entity), ]
NLP_data_entities$token_entity=gsub(pattern = "_-_",replacement = "-",x = NLP_data_entities$token_entity,ignore.case = T,perl = T)
NLP_data_entities$token_entity=gsub(pattern = "_[[:space:]]{1,}_",replacement = "_",x = NLP_data_entities$token_entity,ignore.case = T,perl = T)
#NLP_data_entities=NLP_data_entities[nchar(NLP_data_entities$token_entity)>=3, ]#
NLP_data_entities$token_entity = tolower(NLP_data_entities$token_entity)
NLP_data_entities$token_entity=gsub("’",replacement = "'",x = NLP_data_entities$token_entity)
NLP_data_entities$token_entity=gsub("'_",replacement = "'",x = NLP_data_entities$token_entity)
NLP_data_entities$token_entity=gsub("^[[:punct:]]{1,}|[[:punct:]]{1,}$",replacement = "",x = NLP_data_entities$token_entity)
NLP_data_entities$token_entity=gsub("[[:space:]]{2,}",replacement = " ",x = NLP_data_entities$token_entity)
NLP_data_entities$token_entity=gsub("^[_]{1,}|[_]{1,}$",replacement = "",x = NLP_data_entities$token_entity)
NLP_data_entities$token_entity=gsub("&",replacement = "&",x = NLP_data_entities$token_entity)
saveRDS(NLP_data_entities,paste0("NLP_data_entities_final.RDS"))
#NLP_data_entities=readRDS(paste0("NLP_data_entities_final_en_",Sys.Date(),".RDS"))
source_nodes= NLP_data_entities %>%
dplyr::select(doc_id,
sentence_id,
token_0 = token_entity,
token_0_id=token_id,
pos_0=pos,
dep_rel_01=dep_rel,
token_1_id=head_token_id
)
target_nodes= NLP_data_entities %>% dplyr::select(doc_id,
sentence_id,
pos_1=pos,
token_1 = token_entity,
token_1_id=token_id,
#lemma1 = lemma_nounphrase,
dep_rel_12=dep_rel,
token_2_id=head_token_id
)
net_data=dplyr::left_join(
source_nodes[source_nodes$dep_rel_01!="ROOT",] ,
target_nodes,
by = join_by(
doc_id,
sentence_id,
token_1_id == token_1_id
)
)
#net_data=net_data[net_data$dep_rel!="ROOT",]
net_data=net_data[!is.na(net_data$token_0) & !is.na(net_data$token_1),]
#net_data[,c("word_source","word_target")]
net_data=net_data[!(net_data$token_0 == net_data$token_1),]#remove loops
net_data$token_id=paste(net_data$token_0_id,"|",net_data$token_0_id)#paste("[",net_data$token_0_id,"|",net_data$token_0_id,"]")
net_data$id_up_01=net_data$token_0_id<net_data$token_1_id
net_data$edge_01=NA
net_data$edge_01[net_data$id_up_01]=paste(net_data$token_0[net_data$id_up_01],net_data$token_1[net_data$id_up_01],sep = "->")
net_data$edge_01[!net_data$id_up_01]=paste(net_data$token_1[!net_data$id_up_01],net_data$token_0[!net_data$id_up_01],sep = "->")
temp_data=left_join(net_data,data_clean[,c("doc_id","user_id","user_image")],by = join_by("doc_id"=="doc_id"))
write_csv(temp_data,paste0("deprel_dataedges_final.csv"))
saveRDS(net_data,paste0("net_data.RDS"))
NLP_data_entities$refdoc_id=NLP_data_entities$doc_id
NLP_data_entities$doc_id= paste(NLP_data_entities$doc_id, NLP_data_entities$sentence_id,sep = "_")
quanteda_unigrams=quanteda::as.tokens(NLP_data_entities %>% dplyr::select(doc_id, sentence_id, token_id, token = token_entity)) %>% quanteda::tokens_tolower()
net_data$refdoc_id=net_data$doc_id
net_data$doc_id= paste(net_data$doc_id, net_data$sentence_id,sep = "_")
quanteda_associations=quanteda::as.tokens(net_data %>% dplyr::select(doc_id, sentence_id, token_id, token = edge_01)) %>% quanteda::tokens_tolower()
#BUILD DFMs
#unigrams_dfm =quanteda_unigrams[names(quanteda_associations)]%>% quanteda::dfm()
unigrams_dfm =quanteda_unigrams%>% quanteda::dfm()
unigrams_dfm = unigrams_dfm %>%
dfm_group(.,
groups = gsub("_[0-9]{1,}$",
"",
x = docnames(unigrams_dfm)))
####gsub("_[0-9]{1,}$","", x = docnames(unigrams_dfm)
associations_dfm= quanteda_associations %>% quanteda::dfm()
# colnames(associations_dfm)= gsub(pattern = "<-",
# replacement ="->",x = colnames(associations_dfm))
#ncol(associations_dfm)
#associations_dfm =associations_dfm %>% quanteda::dfm_compress(.,margin = "features")
#ncol(associations_dfm)
associations_dfm= associations_dfm %>%
dfm_group(.,
groups = gsub("_[0-9]{1,}$",
"",
x = docnames(associations_dfm)))
nrow(unigrams_dfm)
nrow(associations_dfm)
saveRDS(unigrams_dfm,paste0("raw_unigrams_dfm_tweets.RDS"))
saveRDS(associations_dfm,paste0("raw_associations_dfm_tweets.RDS"))
table(colSums(associations_dfm)>=2)
table(colSums(unigrams_dfm)>=2)
# unigrams_dfm=unigrams_dfm_bkp
# associations_dfm=associations_dfm_bkp
unigrams_dfm_bkp=unigrams_dfm
associations_dfm_bkp=associations_dfm
# min_freq_uni=round((1/100000)*nrow(data))#once every ten thousand sentences
# min_freq_bi=round((1/100000)*nrow(data))#once every ten thousand sentences
# unigrams_dfm=dfm_trim(unigrams_dfm, min_termfreq = min_freq_uni, docfreq_type = "count")
# associations_dfm=dfm_trim(associations_dfm, min_termfreq = min_freq_bi, docfreq_type = "count")#1 ogni 100 doc
stopwords=c(quanteda::stopwords(language = "fr"),"re","rt","na","si","i",c(1:9))
#unigram_counts=data.frame(term=featnames(unigrams_dfm), freq=colSums(unigrams_dfm))
#unigram_counts[!(unigram_counts$term %in% stopwords),]%>% View()
unigrams_dfm=quanteda::dfm_select(unigrams_dfm,
selection = c("remove"),
valuetype="fixed",
pattern=stopwords)
stopwords_regex=paste(paste("->",stopwords,"$",sep = "",collapse = "|"),paste("<-",stopwords,"$",sep = "",collapse = "|"),paste("^",stopwords,"->",sep = "",collapse = "|"),paste("^",stopwords,"<-",sep = "",collapse = "|"),sep = "|")
associations_dfm=quanteda::dfm_select(associations_dfm,
selection = c("remove"),
valuetype="regex",
pattern=stopwords_regex)
#colnames(associations_dfm)=gsub(pattern = "[<][-]","->",x = colnames(associations_dfm))
#associations_dfm=dfm_compress(associations_dfm,margin ="features")
rownames(data_clean)=data_clean$doc_id
docvars(unigrams_dfm)=data_clean[docnames(unigrams_dfm),]
docvars(associations_dfm)=data_clean[docnames(associations_dfm),]
unigrams_df= tibble(name=featnames(unigrams_dfm), frequency=colSums(unigrams_dfm ))
unigrams_df%>% wordcloud2::wordcloud2(.) #plot wordcloud of unigrams (words)
associations_df= tibble(name=featnames(associations_dfm), frequency=colSums(associations_dfm ) )
associations_df%>% wordcloud2::wordcloud2(.) #plot wordcloud of dep.rel (word associations)
docnames(associations_dfm)[which(!(docnames(associations_dfm) %in% rownames(data_clean)))]
intersection_unigrams_dfm=unigrams_dfm[docnames(associations_dfm),]
nrow(intersection_unigrams_dfm)
ncol(intersection_unigrams_dfm)
identical(docnames(associations_dfm),docnames(intersection_unigrams_dfm))
dfm_intersection = cbind(intersection_unigrams_dfm, associations_dfm)
#docvars(dfm_intersection)
colnames(dfm_intersection)=gsub(pattern = "^[[:punct:]_]{1,}|[[:punct:]_]{1,}$",replacement = "",x = featnames(dfm_intersection),perl = T)
dfm_intersection=dfm_compress(dfm_intersection,margin = "features")
docvars(dfm_intersection)=data_clean[docnames(dfm_intersection),]
saveRDS(dfm_intersection ,paste0("dfm_intersection.RDS"))
complement_unigrams_dfm=unigrams_dfm[!(docnames(unigrams_dfm) %in% docnames(associations_dfm)),]
nrow(complement_unigrams_dfm)
ncol(complement_unigrams_dfm)
complement_associations_dfm=Matrix::Matrix(data=0, nrow=nrow(complement_unigrams_dfm), ncol=ncol(associations_dfm), byrow=FALSE, dimnames=list(docnames(complement_unigrams_dfm),featnames(associations_dfm)), sparse = TRUE, doDiag = TRUE, forceCheck = T) %>% as.dfm()
nrow(complement_associations_dfm)
ncol(complement_associations_dfm)
identical(docnames(complement_associations_dfm),docnames(complement_unigrams_dfm))
dfm_complement= cbind(complement_unigrams_dfm, complement_associations_dfm)
docvars(dfm_complement)=docvars(complement_unigrams_dfm)
#### COMBINING DFMs ####
identical(featnames(dfm_intersection),featnames(dfm_complement))# TRUE is good !
intersect(docnames(dfm_intersection),docnames(dfm_complement)) #character(0) is good!
#dfm= rbind(dfm_intersection, dfm_complement[,featnames(dfm_intersection)])
dfm= rbind(dfm_intersection, dfm_complement)
#docvars(dfm)=data[docnames(dfm),]
docvars(dfm)=data_clean[gsub("_[0-9]{1,}$","", x = docnames(dfm)),]
saveRDS(dfm ,paste0("dfm_final.RDS"))
# Plot Dep.Rel network filtered by word or two words expression
#### Word egonetwork #####
### function to build dep rel network
egonet2=function(filter_term, filter_term2="",filter=0.5, save=TRUE, dfm_=dfm,max_edges=500,max_nodes=2000, ...){
library(visNetwork)
color.gradient_2 = function(x,
colors = c("lightgray", "darkgray"),
colsteps = 1000) {
return(colorRampPalette(colors) (colsteps) [findInterval(x, seq(0, max(bigrams_edges$weight), length.out =
colsteps))])
}
#filter = 0.5
#filter_term="macron"
#filter_term2=""
if(filter_term2!="") {
keep = tolower(paste(
paste(gsub(pattern = " ",replacement = "_",paste0("^",filter_term)),gsub(pattern = " ",replacement = "",filter_term),gsub(pattern = " ",replacement = "[-][>]",paste0(filter_term,"$")),sep = "$|^"),
paste(gsub(pattern = " ",replacement = "_",paste0("_",filter_term)),gsub(pattern = " ",replacement = "",filter_term),gsub(pattern = " ",replacement = "[-][>]",paste0(filter_term,"$")),sep = "$|_"),
paste(gsub(pattern = " ",replacement = "_",paste0("[-][>]",filter_term)),gsub(pattern = " ",replacement = "",filter_term),gsub(pattern = " ",replacement = "[-][>]",paste0(filter_term,"$")),sep = "$|[-][>]"),sep = "|"))
keep2 = tolower(paste(
paste(gsub(pattern = " ",replacement = "_",paste0("^",filter_term2)),gsub(pattern = " ",replacement = "",filter_term2),gsub(pattern = " ",replacement = "[-][>]",paste0(filter_term2,"$")),sep = "$|^"),
paste(gsub(pattern = " ",replacement = "_",paste0("_",filter_term2)),gsub(pattern = " ",replacement = "",filter_term2),gsub(pattern = " ",replacement = "[-][>]",paste0(filter_term2,"$")),sep = "$|_"),
paste(gsub(pattern = " ",replacement = "_",paste0("[-][>]",filter_term2)),gsub(pattern = " ",replacement = "",filter_term2),gsub(pattern = " ",replacement = "[-][>]",paste0(filter_term2,"$")),sep = "$|[-][>]"),sep = "|"))
keep_docs= intersect(names(which(quanteda::dfm_select(dfm_, selection = c("keep"), valuetype="regex", pattern=keep) %>% rowSums()>=1)),
names(which(quanteda::dfm_select(dfm_, selection = c("keep"), valuetype="regex", pattern=keep2) %>% rowSums()>=1)))
dfm_temp =dfm[keep_docs,]
dfm_temp =dfm_temp[,colSums(dfm_temp)>=1]
matched=featnames(dfm_temp)[grepl(pattern = keep,x = featnames(dfm_temp)) |grepl(pattern = keep2,x = featnames(dfm_temp)) ]
}else{
keep = tolower(paste(
paste(gsub(pattern = " ",replacement = "_",paste0("^",filter_term)),gsub(pattern = " ",replacement = "",filter_term),gsub(pattern = " ",replacement = "[-][>]",paste0(filter_term,"$")),sep = "$|^"),
paste(gsub(pattern = " ",replacement = "_",paste0("_",filter_term)),gsub(pattern = " ",replacement = "",filter_term),gsub(pattern = " ",replacement = "[-][>]",paste0(filter_term,"$")),sep = "$|_"),
paste(gsub(pattern = " ",replacement = "_",paste0("[-][>]",filter_term)),gsub(pattern = " ",replacement = "",filter_term),gsub(pattern = " ",replacement = "[-][>]",paste0(filter_term,"$")),sep = "$|[-][>]"),sep = "|"))
keep_docs= names(which(quanteda::dfm_select(dfm_, selection = c("keep"), valuetype="regex", pattern=keep) %>% rowSums()>=1))
dfm_temp =dfm_[keep_docs,]
dfm_temp =dfm_temp[,colSums(dfm_temp)>=1]
matched=featnames(dfm_temp)[grepl(pattern = keep,x = featnames(dfm_temp))]
}
#rownames(data)=data$textID
#p_threshold=.00001
unigrams = colnames(dfm_temp)[!grepl("^.{1,}->{1}.{1,}$", colnames(dfm_temp))]
unigrams_probs = colSums(dfm_temp)[!grepl("^.{1,}->{1}.{1,}$", colnames(dfm_temp))]
unigrams_probs = unigrams_probs / sum(unigrams_probs)
names(unigrams_probs) = unigrams
unigrams_probs=sort(unigrams_probs,decreasing = T)
#filter=0.5
p_threshold = quantile(unigrams_probs, probs = filter)
bigrams_nodes = data.frame(id = names(unigrams_probs)[(unigrams_probs >= p_threshold) |
(names(unigrams_probs) %in% matched)],
label = names(unigrams_probs)[(unigrams_probs >= p_threshold) |
(names(unigrams_probs) %in% matched)],
stringsAsFactors = F)
rownames(bigrams_nodes)=bigrams_nodes$label
bigrams_nodes$value = unigrams_probs[rownames(bigrams_nodes)]
bigrams = colnames(dfm_temp)[grepl("^.{1,}->{1}.{1,}$", colnames(dfm_temp))]
bigrams_probs = colSums(dfm_temp)[grepl("^.{1,}->{1}.{1,}$", colnames(dfm_temp))]
bigrams_probs = bigrams_probs / sum(bigrams_probs)
p_threshold_edges = quantile(bigrams_probs, probs = filter)
names(bigrams_probs) = bigrams
bigrams_probs=sort(bigrams_probs,decreasing = T)
bigrams_edges = data.frame(
from = gsub("[-]{1}[>]{1}.{1,}$", "", x = names(bigrams_probs) ),
to = gsub("^.{1,}[-]{1}[>]{1}", "", x = names(bigrams_probs) ),
weight = bigrams_probs,
edge_name=names(bigrams_probs),
smooth.enabled = T,
smooth.roundness = 0.50,
stringsAsFactors = F
)
bigrams_edges = bigrams_edges[(bigrams_edges$from %in% bigrams_nodes$id) &
(bigrams_edges$to %in% bigrams_nodes$id) &
bigrams_edges$weight >= p_threshold_edges, ]
bigrams_edges$title = paste(
"From-To: '",
bigrams_edges$from,
"'-'",
bigrams_edges$to,
"', Prob: ",
round(bigrams_edges$weight, 6),
sep = ""
)
#bigrams_nodes = bigrams_nodes[bigrams_nodes$id %in% c(bigrams_edges$from, bigrams_edges$to),] #remove isolated nodes
bigrams_edges = bigrams_edges[!diag(sapply(bigrams_edges$from, `==`, bigrams_edges$to)), ] #remove loops
if(nrow(bigrams_nodes)>max_nodes){
bigrams_nodes=bigrams_nodes[(1:nrow(bigrams_nodes) %in% 1:max_nodes) | (bigrams_nodes$label %in% matched), ]
bigrams_edges = bigrams_edges[(bigrams_edges$from %in% bigrams_nodes$id) &
(bigrams_edges$to %in% bigrams_nodes$id), ]
}
if(nrow(bigrams_edges)>max_edges){
bigrams_edges=bigrams_edges[(1:nrow(bigrams_edges) %in% 1:max_edges) | (bigrams_edges$from %in% matched) | (bigrams_edges$to %in% matched)|(bigrams_edges$edge_name %in% matched), ]
bigrams_nodes = bigrams_nodes[bigrams_nodes$id %in% c(bigrams_edges$from, bigrams_edges$to,matched),]
}
#topic_bigrams_edges$weight
bigrams_nodes$font.color= "black"
bigrams_nodes$shape = "text"#dot
bigrams_nodes$font.color[bigrams_nodes$id %in% matched]="darkred"
bigrams_edges$color = color.gradient_2(bigrams_edges$weight)
bigrams_nodes$title = paste(
"Term: '",
bigrams_nodes$label,
"', Prob: ",
as.character(round(bigrams_nodes$value,digits = 5)),
sep = ""
)
bigrams_edges$width = (bigrams_edges$weight / max(bigrams_edges$weight)) *
5
Network = visNetwork::visNetwork(
main = list(text =paste('<b>Words association network for: ',filter_term,ifelse(filter_term2=="","", paste0(" & ",filter_term2)),"</b>",sep = ""),
style = "font-size:25px;text-align:center;"),
submain = list(text = paste('<b>(nodes & edges percentile filtering threshold = ',filter,'</b>)<br><i>Network based on aggregated and filtered words dependency tree graphs of X bios/tweets containing the chosen term(s).<br>A total of ',nrow(dfm_temp),' documents matched the selected term(s).</i>',sep = ""),
style = "font-size:15px;text-align:center;"),
footer = paste('SoMe4Dem project - figure based on aggregated data from X - version:',Sys.Date()),
height = "1400px",
width = "100%",
nodes = bigrams_nodes,
edges = bigrams_edges
) %>% visEdges(arrows = "to", smooth = T) %>%
visNodes(scaling = list(label = list(min = 20, max = 100))) %>%
visPhysics(solver = "forceAtlas2Based",stabilization= FALSE,
forceAtlas2Based = list(gravitationalConstant = -50)) %>%
visInteraction(multiselect = TRUE, navigationButtons = TRUE) %>%
visOptions(
manipulation = TRUE,
nodesIdSelection = list(enabled = TRUE),
highlightNearest = list(enabled = T, degree = 2, hover = T)
)
#%>% visHierarchicalLayout(direction = "LR", levelSeparation = 300,edgeMinimization = F,treeSpacing = 300,blockShifting = T)
#Network %>%
if(save==T){
visSave(Network, paste0(filter_term,ifelse(filter_term2=="","_", paste0("_AND_",filter_term2,"_")),Sys.Date(),"_filter",filter,".html"), selfcontained = TRUE, background = "white")
}
Network
}
### egonet plot
egonet2(filter_term = "[query]",max_nodes = 200,max_edges = 1000,filter = 0)# filter is a percentile filter with a range from 0 to 1