jwijffels
commented
4 years ago That's funny. I was disturbed a bit first but it does detect it further away. It is because you work on the lemma and cannothad as lemma can which was not part of your negator list
library(udpipe)
x <- "i cannot understand when people say that the noise-cancellation on these is far superior"
anno <- udpipe(x, "english-gum", udpipe_model_repo = "jwijffels/udpipe.models.ud.2.3")
scores <- txt_sentiment(x = anno,
term = "lemma",
polarity_terms = data.frame(term = c("annoy", "like", "painful", "superior"),
polarity = c(-1, 1, -1, 1)),
polarity_negators = c("can"),
polarity_amplifiers = c("pretty", "many", "really", "whatsoever"),
polarity_deamplifiers = c("slightly", "somewhat"),
constrain = TRUE, n_before = 20,
n_after = 20, amplifier_weight = .8)
View(scores$data)
token lemma upos xpos head_token_id dep_rel deps polarity sentiment_polarity
1 i us NOUN NNS 3 nsubj <NA> NA NA
2 cannot can AUX MD 3 aux <NA> NA NA
3 understand understand VERB VB 0 root <NA> NA NA
4 when when SCONJ WRB 6 advmod <NA> NA NA
5 people people NOUN NNS 6 nsubj <NA> NA NA
6 say say VERB VBP 3 advcl <NA> NA NA
7 that that SCONJ IN 14 mark <NA> NA NA
8 the the DET DT 9 det <NA> NA NA
9 noise-cancellation noise-cancellation NOUN NN 14 nsubj <NA> NA NA
10 on on ADP IN 11 case <NA> NA NA
11 these these PRON DT 9 nmod <NA> NA NA
12 is be AUX VBZ 14 cop <NA> NA NA
13 far far ADV RB 14 advmod <NA> NA NA
14 superior superior ADJ JJ 6 ccomp <NA> 1 -1
PeteTat
The polarity negator detects the word if it is next to term positive or negative however it does not detect if it is further away.
`library("udpipe")
dmodel_english <- udpipe_load_model(file = "./english-gum-ud-2.3-181115.udpipe")
x <- c("i cannot understand when people say that the noise-cancellation on these is far superior")
anno <- udpipe(x, dmodel_english)
anno <- data.frame(anno)
scores <- txt_sentiment(x = anno, term = "lemma", polarity_terms = data.frame(term = c("annoy", "like", "painful", "superior"), polarity = c(-1, 1, -1, 1)), polarity_negators = c("not", "neither", "cannot"), polarity_amplifiers = c("pretty", "many", "really", "whatsoever"), polarity_deamplifiers = c("slightly", "somewhat"), constrain = TRUE, n_before = 15, n_after = 15, amplifier_weight = .8) scores$overall`