In PolMine/polmineR.topics: tools for topicmodelling

Initialization

Settings

instancefile <- tempfile() # Use findable path
modelfile <- tempfile() # Use findable path

The memory available for the Java Virtual Machine (JVM) to be used needs to be defined before creating the JVM

options(java.parameters = "-Xmx4g") # Sufficient for larger data

Load libraries

library("polmineR")
library(topicanalysis)
library(mallet) # Includes mallet jars
library(rJava)
library(data.table)

Corpus data

Create document-level data

use("polmineR")
coi <- "GERMAPARLMINI"
speeches <- as.speeches(coi, s_attribute_name = "speaker")

R-side preprocessing

Keep only documents with a minimum length.

doc_min_length <- 100L
dt <- as.data.table(summary(speeches))
speeches <- speeches[[ dt[size >= doc_min_length][["name"]] ]]

Create instance list

instance_list <- mallet_make_instance_list(speeches)

Implicitly, the mallet_make_instance_list uses stopwords of the tm package.

Estimate topic model

Starting to estimate the topic model at: r (format(started <- Sys.time(), format = "%T"))

lda <- .jnew("cc/mallet/topics/ParallelTopicModel", 25L, 5.1, 0.1)
lda$addInstances(instance_list)
lda$setNumThreads(1L)
lda$setTopicDisplay(50L, 10L)
lda$setNumIterations(2000L)
lda$estimate()

Finished computation at r (format(finished <- Sys.time(), "%T")) (total time: r format(Sys.time() - started, format = "%T")).

Save topic model

lda$write(rJava::.jnew("java/io/File", modelfile))

PolMine/polmineR.topics documentation built on March 6, 2020, 6:03 p.m.