aggregate.sentiment: Aggregate textual sentiment across sentences, documents and...

In sentometrics: An Integrated Framework for Textual Sentiment Time Series Aggregation and Prediction

Description

Aggregates textual sentiment scores at sentence- or document-level into a panel of textual sentiment measures. Can also be used to aggregate sentence-level sentiment scores into document-level sentiment scores. This function is called within the sento_measures function.

Usage

## S3 method for class 'sentiment'
aggregate(x, ctr, do.full = TRUE, ...)

Arguments

x	a sentiment object created using compute_sentiment (from a sento_corpus object) or using as.sentiment.
ctr	output from a ctr_agg call. The howWithin and nCore elements are ignored.
do.full	if do.full = TRUE (by default), does entire aggregation up to a sento_measures object, else only goes from sentence-level to document-level. Ignored if no "sentence_id" column in sentiment input object.
...	not used.

Value

A document-level sentiment object or a fully aggregated sento_measures object.

Author(s)

Samuel Borms, Keven Bluteau

See Also

compute_sentiment, ctr_agg, sento_measures

Examples

set.seed(505)

data("usnews", package = "sentometrics")
data("list_lexicons", package = "sentometrics")
data("list_valence_shifters", package = "sentometrics")

# computation of sentiment
corpus <- sento_corpus(corpusdf = usnews)
corpusSample <- quanteda::corpus_sample(corpus, size = 500)
l1 <- sento_lexicons(list_lexicons[c("LM_en", "HENRY_en")],
                     list_valence_shifters[["en"]])
l2 <- sento_lexicons(list_lexicons[c("LM_en", "HENRY_en")],
                     list_valence_shifters[["en"]][, c("x", "t")])
sent1 <- compute_sentiment(corpusSample, l1, how = "counts")
sent2 <- compute_sentiment(corpusSample, l2, do.sentence = TRUE)
sent3 <- compute_sentiment(as.character(corpusSample), l2,
                           do.sentence = TRUE)
ctr <- ctr_agg(howTime = c("linear"), by = "year", lag = 3)

# aggregate into sentiment measures
sm1 <- aggregate(sent1, ctr)
sm2 <- aggregate(sent2, ctr)

# two-step aggregation (first into document-level sentiment)
sd2 <- aggregate(sent2, ctr, do.full = FALSE)
sm3 <- aggregate(sd2, ctr)

# aggregation of a sentiment data.table
cols <- c("word_count", names(l2)[-length(l2)])
sd3 <- sent3[, lapply(.SD, sum), by = "id", .SDcols = cols]

sentometrics documentation built on Aug. 18, 2021, 9:06 a.m.