library(udpipe)
library(tidyverse)
library(tidygraph)
library(ggraph)
library(httr)
#Загружаем данные
set_config(config(ssl_verifypeer = 0L))
url <- "https://github.com/locusclassicus/text_analysis_2024/raw/main/files/bg_latinpipe.conllu"
caesar <- udpipe_read_conllu(url)
#расчет существительных
cooc_data <- cooccurrence(
caesar |> filter(upos == "NOUN"),
term = "lemma",
group = c("doc_id", "sentence_id")
) |>
as_tibble() |>
filter(cooc > 40)
#создаем граф
graph_data <- as_tbl_graph(cooc_data, directed = FALSE) |>
mutate(
degree = centrality_degree(),
group = ifelse(degree > 3, "Частые", "Редкие"),
component = group_components()
) |>
filter(component == 1)
#визуализация
set.seed(21092024)
ggraph(graph_data, layout = "fr") +
geom_edge_arc(aes(edge_width = cooc),
color = "#2C3E50",
alpha = 0.8,
strength = 0.1,
arrow = arrow(angle = 25,
length = unit(0.2, "cm"),
ends = "both",
type = "closed"),
start_cap = circle(10, "mm"),
end_cap = circle(10, "mm"),
show.legend = FALSE) +
geom_node_point(aes(size = degree, fill = group),
shape = 21,
color = "black",
stroke = 1.2) +
geom_node_text(aes(label = name),
nudge_y = 1.2,
size = 4.5,
fontface = "bold") +
#прописываем цвета
scale_fill_manual(values = c("Частые" = "#0000FF", "Редкие" = "#FF0000")) +
scale_size(range = c(4, 10), guide = "none") +
scale_edge_width(range = c(0.8, 3.5)) +
scale_y_continuous(expand = expansion(mult = c(0.2, 0.3))) +
scale_x_continuous(expand = expansion(mult = 0.2)) +
theme_graph(base_family = "sans") +
theme(legend.position = "bottom") +
labs(
title = "Сеть связей существительных",
)
Стоит отметить: при рендеринге почему-то не отображаются все связи с необходимыми “стрелками”. Для ясности прикрепляю финальный вариант.
